Dementia Attributable Healthcare Utilizations in the Caribbean versus United States

INTRODUCTION

Dementia is a highly prevalent and expensive chronic condition for health systems, patients, and their caregivers, of which Alzheimer’s disease is the most common. There are an estimated 55 million people aged 65 and older living with Alzheimer’s disease dementia globally in 2021 with an estimated 10 million new cases every year [1]. Individuals with dementia incur higher health care spending than those without dementia [2], partly because utilization of health care services for other serious medical conditions is strongly affected by the presence or absence of dementia, such as coronary artery disease, diabetes, stroke, or cancer [1, 3]. Specifically, dementia has been associated with more intensive healthcare utilizations including hospitalizations, emergency department visits, and life-sustaining procedures [4, 5]. Dementia, of which Alzheimer’s disease represents the majority of cases, thus represents a unique burden to healthcare systems worldwide.

Compared to high-income countries such as the US, relatively limited evidence exists regarding dementia attributable healthcare utilization in developing countries, in particular the Caribbean region. Caribbean Hispanics, especially those of Cuban, Puerto Rican, and Dominican descent, are the second largest Hispanic group in the US [6]. Hispanics in general have been identified as a high-risk minority population for dementia: in the US, the prevalence of Alzheimer’s disease and related dementias (ADRD) among Hispanics aged 65 and older is nearly 17%, which is over two times higher than non-Hispanic Whites of the same age [7]. Hispanics with ADRD are found to have higher rates of intensive healthcare utilization, including hospitalizations and emergency department visits than Whites [7, 8]. These have been largely attributed to socioeconomic and cultural factors including language barriers, lack of access to high-quality healthcare and misconception of dementia as a normal part of aging [7]. However, existing literature on dementia-related healthcare burdens among Hispanics are primarily US based, and focus mostly on Mexican Hispanics, with very few studies focusing on Caribbean Hispanics.

An improved understanding of dementia attributable healthcare utilization in the Caribbean region could offer unique insights to dementia-related disparities and economic burden, both in the US and globally. While certain factors that impede equitable access to quality healthcare in the US, such as language barriers, are likely less relevant in their home countries, Caribbean-dwelling Hispanics face additional challenges for dementia care. A general lack of social care infrastructure, such as homecare and residential care, could mean that dementia patients in the Caribbean region and many other developing countries may rely more heavily on medical care in addition to informal care [9, 10]. In fact, medical care is estimated to account for over a quarter of societal cost of dementia in the Caribbean islands, ranging from 25% in Puerto Rico to 35% in Dominican Republic, compared to 18% in the US [9]. This is further exacerbated by inadequate insurance coverage particularly in the Dominican Republic [10, 11] (see Supplementary Table 7 for an overview of country-level statistics, including out-of-pocket share of healthcare expenditure by country/region). In comparison, medical care is free in Cuba [12]. Adults aged 65 and over in the US (including Puerto Rico) have near-universal insurance coverage through Medicare, although prior literature has documented generally worse quality of healthcare experienced by Medicare enrollees in Puerto Rico compared to their US mainland counterparts [13, 14]. Finally, delayed care due to misconception of dementia as a normal part of aging could also contribute to intensive healthcare utilization among Caribbean-dwelling Hispanics as among US Hispanics.

In this study, we examined healthcare utilization associated with dementia among older adults in both the US (by race and ethnicity) and Caribbean region including Cuba, Dominican Republic (DR), and Puerto Rico (PR). Specifically, we examined the extent to which new dementia onset (in past two years) and ongoing dementia (present two years earlier) may be associated with elevated hospital admissions and outpatient physician visits in these diverse populations. A few factors may drive changes in these different dimensions of healthcare utilization in response to new and ongoing dementia. On the one hand, delayed diagnosis of dementia (either due to lack of adequate healthcare access or lack of patient awareness of the significance of early symptoms) could lead to more intensive utilization, especially hospitalization, at both the onset of dementia and even among patients with ongoing dementia [15]. On the other hand, to the extent that there is timely detection of dementia by healthcare providers and it is being managed in the outpatient setting, there are likely increased physician visits associated with both incident and ongoing dementias [15, 16]. Comparing dementia attributable utilization between older adults in the US and the Caribbean region not only provides an overall assessment of the burden of dementia on these diverse healthcare systems, but also provides suggestive evidence on the relative importance of potential contributing factors to dementia-related disparities by context and race/ethnicity. Comparisons with the US populations using harmonized data and analyses allow us to better benchmark the findings on the Caribbean populations and also draw comparisons with the existing US based literature.

To this end, we employed and harmonized two population-based longitudinal survey data sources on older adults in the US and Caribbean sites (Cuba, DR, and PR), namely, the Health and Retirement Study and the 10/66 Dementia Research Group Cohort Study. Both data sources contain measures on healthcare utilization (namely hospital admissions and physician visits), dementia, as well as key socio-demographics and comorbidities that are important to account for in order to assess the independent associations between dementia and healthcare utilization across contexts and populations.

MATERIALS AND METHODS

RESULTS

Mean (SD) age in the pooled Caribbean sample was 77.6 (6.5), slightly lower than 79.5 (7.3) in the pooled US sample (p < 0.001), with similar age across subsamples (Table 1). The proportion of females was higher in the Caribbean sample than in the US sample (68% versus 57%; p < 0.001). Hypertension (prevalence at Wave 2:60% in the Caribbean sample, 67% in the US sample) and arthritis (55% in Caribbean, 73% in the US) were the two most prevalent chronic conditions in both samples. Within a three-month period, 3.4% of the Caribbean sample spent at least one night in a hospital, and 51% had at least one physician visit, with an average number of 0.4 hospital nights and 1.6 physician visits. Within a two-year period, 34% of the US sample spent at least one night in a hospital, and 93% had any physician visit, with an average number of 2.9 hospital nights and 12.4 physician visits. Dementia prevalence increased from 7.0% in Wave 1 to 15.2% in Wave 2 in the Caribbean sample, and from 4.9% to 10.9% in the US sample, more than doubled in both samples. At Wave 2, 9.4% of the Caribbean and 7.4% of the US samples were classified as having incident dementia, and 5.7% and 3.5% were classified as ongoing dementia. Dementia prevalence in both waves was the highest in DR among the three Caribbean countries/islands, and both Blacks and Hispanics had higher dementia prevalence than Whites in the US.

We also showed the characteristics of our study sample in both waves compared to all respondents in each data source in the first wave in Supplementary Tables 2 and 3. About 63% (4,265 out 6,737) of all participants in Wave 1 of the 10/66 data were followed up successfully in Wave 2 and included in our sample, whereas the remaining were not present in Wave 2 either due to mortality attrition or lost to follow-up. The analogous statistic in the HRS data were 68% (6,854 out of 10,087). Compared to the full sample in Wave 1, our study samples in Wave 1 had generally lower healthcare utilization and much lower dementia prevalence, in both data sources.

Figures 1–4 show graphically the adjusted results from multivariable regressions, for each of the four dependent variables. Incident dementia was associated with a statistically significant difference in the risk of hospitalization relative to individuals without dementia in the Caribbean sample (RR: 1.71, 95% CI 0.96–2.46, p = 0.018), with insignificant association for ongoing dementia (RR: 1.30, 95% CI 0.43–2.18, p = 0.443) (Fig. 1). This pattern was similar across Caribbean sites except Cuba and was more prominent among men (Supplementary Table 4A). In comparison, both dementia statuses increased the risk of hospitalization in the US sample by a factor of about 1.3 (RR: ongoing dementia 1.27, 95% CI 1.07–1.47, p = 0.005; incident dementia 1.31, 95% CI 1.17–1.45; p < 0.001). Results stratified by race and ethnicity group showed that the relationship between ongoing dementia and risk of hospitalization was only significant among Whites (RR: 1.35, 95% CI 1.09–1.62; p = 0.007), while the relationship between incident dementia and risk of hospitalization was significant across all racial and ethnic groups (Supplementary Table 4B).

OPEN IN VIEWER

Fig. 1

Relative risk ratios of spending any night in the hospital associated with ongoing dementia and incident dementia relative to no dementia, in the Caribbean and US samples. The height of each bar represents the value of the estimated relative risk ratio from a logistic regression with any hospital night as the dependent variable and ongoing and incident dementia as the independent variables, controlling for age, education, income, marital status, and chronic conditions as shown in Table 1. The whiskers represent the 95% confidence intervals of the relative risk ratios. The two bars on the left show results estimated from a single regression on the full Caribbean sample. The two bars on the right show results estimated from a single regression on the full US sample.

OPEN IN VIEWER

Fig. 2

Incident rate ratios of number of hospital nights associated with ongoing dementia and incident dementia relative to no dementia, in the Caribbean and US samples. The height of each bar represents the value of the estimated incident rate ratio from a negative binominal regression with number of hospital nights as the dependent variable and ongoing and incident dementia as the independent variables, controlling for age, education, income, marital status, and chronic conditions as shown in Table 1. The whiskers represent the 95% confidence intervals of the incident rate ratios. The two bars on the left show results estimated from a single regression on the full Caribbean sample. The two bars on the right show results estimated from a single regression on the full US sample.

OPEN IN VIEWER

Fig. 3

Relative risk ratios of having any physician visit associated with ongoing dementia and incident dementia relative to no dementia, in the Caribbean and US samples. The height of each bar represents the value of the estimated relative risk ratio from a logistic regression with any physician visit as the dependent variable and ongoing and incident dementia as the independent variables, controlling for age, education, income, marital status, and chronic conditions as shown in Table 1. The whiskers represent the 95% confidence intervals of the relative risk ratios. The two bars on the left show results estimated from a single regression on the full Caribbean sample. The two bars on the right show results estimated from a single regression on the full US sample.

OPEN IN VIEWER

Fig. 4

Incident rate ratios of number of physician visits associated with ongoing dementia and incident dementia relative to no dementia, in the Caribbean and US samples. The height of each bar represents the value of the estimated incident rate ratio from a negative binominal regression with number of physician visits as the dependent variable and ongoing and incident dementia as the independent variables, controlling for age, education, income, marital status, and chronic conditions as shown in Table 1. The whiskers represent the 95% confidence intervals of the incident rate ratios. The two bars on the left show results estimated from a single regression on the full Caribbean sample. The two bars on the right show results estimated from a single regression on the full US sample.

Incident dementia was associated with an increase in the number of hospital nights of over three folds compared to cognitively normal individuals during a three-month period in the Caribbean sample (incidence rate ratio (IRR): 3.43, 95% CI 1.51–7.79, p = 0.003) (Fig. 2). The increase in number of hospital nights attributable to incident dementia was particularly prominent in PR (IRR: 9.11, 95% CI 2.36–25.19, p = 0.001) and among males (IRR: 10.66, 95% CI 3.32–34.25, p < 0.001) (Supplementary Table 5A). On other hand, there was little difference in number of hospital nights between those with ongoing dementia versus no dementia. In DR, however, ongoing dementia was associated with a decrease in number of hospital nights (IRR: 0.01, 95% CI 0.00–0.06, p < 0.001). In the US sample, incident dementia was associated with an approximately two-fold increase in number of hospital nights over a two-year period (IRR: 2.25, 95% CI 1.69–2.99; p < 0.001), and ongoing dementia was associated with an increase in the IRR as well (IRR: 1.64, 95% CI 1.24–2.15; p < 0.001) (Fig. 2 and Supplementary Table 5B). The increase in hospital nights attributed to incident dementia was observed in all racial and ethnic groups and was especially large among Hispanics (IRR: 5.89, 95% CI 2.07–16.80; p < 0.001) (Supplementary Table 5B). Hispanic men appeared to be especially affected by incident dementia (IRR: 5.91, 95% CI 1.69–20.68; p = 0.005), although with overlapping confidence interval with Hispanic women (IRR: 3.03, 95% CI 0.76–12.07; p = 0.115).

Neither incident nor ongoing dementia was associated with a change in the risk of having any physician visits, and this is the case in both the Caribbean and US samples, as well as across subsamples (Fig. 3 and Supplementary Table 6). In the US sample, ongoing dementia was associated with an almost two-fold increase in the number of physician visits during a two-year period (IRR: 1.96, 95% CI 1.25–3.08; p = 0.003) (Fig. 4 and Supplementary Table 7B). This relationship was especially prominent among Blacks (IRR: 2.53, 95% CI 1.43–4.48; p = 0.001). Incident dementia was associated with a 1.34-fold increase in these visits for the full US sample (IRR: 1.34, 95% CI 1.12–1.61; p = 0.002), though it was only significant among Whites (IRR: 1.49, 95% CI 1.21–1.82, p < 0.001). By contrast, there were again no statistically significant differences in number of physician visits for either incident dementia or ongoing dementia in the Caribbean sample or subsamples (Fig. 4 and Supplementary Table 7A). Finally, results from the propensity-score weighted Hispanic subsample in the US (Supplementary Table 8) were very similar to the unweighted results in Column 5, Panel B of Supplementary Tables 4–7.

DISCUSSION

We showed that among Caribbean-dwelling Hispanics, incident dementia was associated with a significant increase in the risk of hospitalization and number of hospital nights. While we did not observe the same pattern for ongoing dementia, it was associated with a decrease in number of hospitalizations in DR. We did not find any relationship between incident or ongoing dementia and physician visits in the Caribbean sample. On the other hand, among the US sample, both incident and ongoing dementia were associated with a higher risk of having any hospital nights as well as an increased number of hospital nights, and both forms of dementia were associated with an increase in the number of physician visits.

Our findings suggest that, compared to healthcare system in the US, dementia related burden on the Caribbean healthcare systems is concentrated in hospital care, and among patients with incident dementia, especially men. This may be driven by a few possible factors. On the healthcare supply side, it is more common for initial dementia diagnosis to made during a short (usually two- to three-day) hospital stay in the Caribbean region as opposed to during an outpatient visit as that in the US. Accordingly, compared to the US, there could be poorer ongoing chronic condition management, including diagnosis and management of dementia in the outpatient setting in the Caribbean countries/islands [27]. This may give rise to acute episodes from worsening of existing conditions as the patient experiences dementia onset.

The general differences in healthcare financing and access across countries and regions could also be contributing to the findings (Supplementary Table 1). Of the four contexts, DR had the lowest percentage of healthcare expenditure as share of Gross Domestic Product (GDP), lowest number of physicians and hospital beds per 1,000 population, and the highest out-of-pocket share of healthcare expenditures, pointing to a lack of insurance coverage and inadequate healthcare access. The fact that we found lower rate of hospitalizations among older adults with ongoing dementia compared to no dementia in DR could be due to lower socioeconomic status and inadequate insurance coverage in the dementia population [11]. Even in Cuba and PR where insurance coverage is presumably universal or near-universal, there could be inadequate management of chronic conditions, which would be consistent with existing evidence for lower quality of care in PR relative to mainland US Medicare [14]. A related possible contributing factor is the lack of diagnostic skills, tools or knowledge of dementia among healthcare providers [28], which is also an issue in the US but could be more prevalent in the Caribbean region. Finally, a general lack of post-acute care and long-term care providers and infrastructures in the Caribbean region could also lead to longer hospital stays [9]. It is noteworthy that we found similar patterns in terms of increased hospital nights in response to incident dementia in the US particularly among Blacks and Hispanics compared with Whites, which likely reflects documented racial and ethnic disparities in both routine healthcare and dementia care in the US [29 –31].

On the other hand, supplier-induced demand or at least inefficient care could be contributing to our finding of higher intensity of hospital and physician care in the US among those with ongoing dementia. Existing literature using data from Sweden suggests that healthcare costs were not necessarily higher among dementia patients four to six years after diagnosis. Given that US had both relatively high numbers of physicians and hospital beds per 1,000 population and a more fragmented healthcare system in terms of sources of financing (Supplementary Table 1), it is possible that patients with dementia in the US were more likely to receive unnecessary or low-value care due to their vulnerability compared to those in the Caribbean region. This is an important topic for future research.

On the patient demand side, the potential misconception of dementia symptoms as a normal part of aging may also prevent dementia patients and their families among the Caribbean-dwelling Hispanic population from seeking timely care. Such perception of dementia has been shown to be particularly prominent among Hispanics compared to Whites in the US [7, 32]. The fact that we found larger increases in number of hospital nights in response to incident dementia among men in the Caribbean sample may be related to generally lower use of healthcare services and less contact with the healthcare system prior to dementia onset relative to women [33]. A direct consequence of the lack of demand for timely management of dementia, which is also affected by the access issue discussed above, is potentially higher utilization of informal care in place of formal care for dementia patients in the Caribbean region relative to the overall US population. Lower labor market opportunity cost has also been documented as a key driver for the large amount of informal care provided by family members among disadvantaged households in the Caribbean and Latin America [34].

Our study has limitations. First, the method for assigning dementia status differed in the Caribbean and US samples due to differences in cognitive assessment information between data sources, which could influence the comparability of our estimates. Prior work mitigated this concern to an extent by showing consistency in results across different dementia classification mechanisms in the HRS data [25]. Second, the healthcare utilization measures in the 10/66 were measured during a relatively short time period compared to HRS (three months versus two years), which could lead to noisy estimates for the Caribbean sample. Third, relatedly, to the extent that there was recall bias, it could affect the quality of self-reported utilization measures differently in the HRS versus 10/66 in that utilizations could be under-reported in HRS especially among those with dementia. Fourth, our main study sample likely captured mostly mild to moderate dementia cases as the most severe dementia cases may have dropped out or deceased between Wave 1 and Wave 2, and this could have contributed to the null results on ongoing dementia in the Caribbean sample, although the US sample had largely similar attrition rate between waves. Finally, we were unable to directly examine informal care utilization attributable to incident and ongoing dementia due to substantial differences in relevant survey questionnaires in the 10/66 and HRS data.

Despite these limitations, our study is the first to compare dementia attributable healthcare utilization between the Caribbean and the US populations by utilizing and harmonizing the best available population-based data sources in these two contexts. Taken together, although our findings do not suggest particularly high dementia attributable healthcare burden in the Caribbean compared to the US, they point to potential deficiencies in outpatient management of ADRD as well as other chronic conditions particularly in the Caribbean and suggest potentially high burden of informal care in the region. One important but little explored question is how quality of life of dementia patients in the Caribbean compared with that in the US by race and ethnicity, given differences in utilization of formal healthcare care. Future studies may explore these research questions by comparing both informal care utilization as well as measures of quality of life between these two populations.