Abstract
We aimed to describe admission rates and outcomes of hospitalized people with Alzheimer’s disease (AD) or senile dementia stratified by type 2 diabetes (T2D) in Spain, 2004 to 2013 (N = 541 858, 143 501 [26.5%] of whom had T2D). We excluded alternative causes of dementia. Hospitalization rates were higher in people with T2D (130.5 vs 91.5 cases/105 people). People older than 84 years and comorbidity increased over time. Crude inhospital mortality (IHM) decreased over time both in people with and without T2D (all P values <.001). Time trend analyses, 2004 to 2013, showed an overall adjusted incidence rate ratio (95% confidence interval [95% CI]) of hospitalization of 1.41 (1.40-1.42) for T2D (men, 1.32 [1.30-1.33]; women, 1.46 [1.45-1.48]). In logistic regression analyses, IHM decreased over time (odds ratio, OR [95% CI] = 0.97 [0.96-0.98]) and T2D was not associated with a higher IHM (OR [95% CI] = 0.99 [0.98-1.01]). In conclusion, admission rates were higher in patients with T2D. Inhospital mortality decreased over time. Diabetes did not predict IHM in patients with AD or senile dementia.
Introduction
The prevalence of type 2 diabetes (T2D) is steadily rising. The number of people with T2D has more than doubled in Spain over the last decade, mainly due to increasing obesity rates and a progressively older population. 1 Alzheimer’s disease (AD) is the most common form of dementia and the sixth leading cause of death in the United States. 2 The relationship between AD and T2D is complex. On one hand, there seems to be an increased risk of developing mild cognitive impairment, dementia, or AD in individuals with T2D. 3,4 On the other hand, recent studies suggest that AD pathology increases susceptibility to diabetes due to impaired insulin signaling in the hypothalamus. 5 Some authors have even gone further in this association, calling AD “type 3 diabetes.” 6 Experimental studies support the fact that abnormalities in insulin and insulin growth factor signaling mechanisms begin early in the course of AD, progress with severity of neurodegeneration, and might therefore have an important role in the pathogenesis of AD. 7
Patients with AD are frequently admitted to the hospital, but dementia itself is usually not the main diagnosis at admission. 8 Infections, dehydration, and diabetes or its complications have been highlighted as common reasons for hospital admission in patients with dementia. 9 Although some authors have reported a high mortality burden attributable to AD, 10,11 other researchers have not found an association between mortality and AD. 12 Patients with dementia living in nursing homes may receive a lower provision of diabetes-related recommended procedures, 13 which might turn into increased risks of clinical events. 14 A recent meta-analysis supports an association between diabetes and mortality in patients with dementia, but the population studied was not restricted to patients admitted to hospital, but also included community-dwelling and nursing home residents. 15
In this study, we used Spanish national hospital discharge data (2004-2013) to examine trends in incidence rates and outcomes of AD or senile dementia among patients with or without T2D. In particular, we focused on patient comorbidities, medical complications, diagnostic and therapeutic procedures, and outcomes, such as inhospital mortality (IHM) and length of hospital stay (LOHS). We aimed to evaluate to what extent T2D can possibly influence the clinical profiles of the hospitalized patients with AD or senile dementia and their clinical courses.
Materials and Methods
We performed a retrospective observational study using the Spanish National Hospital Database (Conjunto Mínimo Básico de Datos [CMBD]), which is managed by the Spanish Ministry of Health, Social Services and Equality. In Spain, the National Health Service provides universal, free medical coverage; however, some citizens also have private insurances. This data set compiles all public and private hospital data, covering more than 95% of hospital admissions. 16 (Due to space constraints, see Table S1 for additional explanations on what the Spanish National Hospital Database consists of, on the exclusion criteria for the study, and on the diagnostic and procedural codes retrieved for the analyses.) We analyzed data collected between January 1, 2004, and December 31, 2013 (10 complete years) for patients aged 70 and older.
The criteria for diseases and procedures were defined according to the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM), which is used in the Spanish CMBD. For the main analyses, we selected admissions for patients whose medical diagnoses included AD and senile dementia, coded as 331.0, 290.0, 290.20, 290.21, and 290.3 in any diagnosis field according to the ICD-9-CM. Had we only included those cases coded as 331.0, we would have then missed many cases of AD. To further improve specificity, we excluded patients who could potentially have alternative causes of dementia (see Table S1). We grouped admissions by diabetes status as follows: type 1 diabetes (T1D, ICD-9-CM codes: 250·x1 and 250·x3), T2D (ICD-9-CM codes: 250·x0 and 250·x2), and no-diabetes. People without diabetes and T1D cases were excluded from the study.
Clinical characteristics included information on overall comorbidity at the time of diagnosis, which was assessed by calculating a modified Charlson comorbidity index (CCI) built after removing the items of diabetes and dementia. 17 Irrespective of the position at the diagnoses coding list, we retrieved data on several additionally prespecified ICD-9-CM codes at discharge (“risk factors”) and procedures (see Table S1). We defined readmission as whether the current admission took place within the 30-day period after a former hospitalization. We estimated the mean LOHS and the IHM as defined by the proportion of patients who died inhospital for each year studied.
Statistical Analyses
In order to assess time trends, we estimated the rates of hospitalization for any reason per 100 000 inhabitants in people with AD or senile dementia, stratified by T2D status. We calculated T2D-specific incidence rates by dividing the number of admissions per year, sex, and age-group by the corresponding number of people living in Spain with T2D in that population group. To estimate the people with T2D in each sex and age-group, we multiplied the estimated prevalence of T2D for each category, obtained from National Health Surveys conducted in 2003 to 2004, 2006 to 2007, 2009 to 2010, and 2011 to 2012,
18
and from data of the
A descriptive statistical analysis was performed for all continuous variables and categories by stratifying admissions in people with AD according to T2D status. Variables are expressed as proportions, means with standard deviations, or medians with interquartile ranges (ie, LOHS). A bivariate analysis of variables according to year was performed using the χ2 test for linear trend (proportions), analysis of variables (means), and Kruskal-Wallis (medians), as appropriate.
In order to test the time trend for rates of admissions and to assess the adjusted effect of diabetes on incidence rates, we fit a Poisson regression model, using year of discharge, sex, age, CCI, and T2D status as independent variables.
To assess the influence of T2D on IHM, we performed logistic regression analyses for the entire population taking IHM as the dependent outcome. We accounted for the independent variables included in the Poisson regression model plus other prespecified potential confounders, such as readmission and codes of different diagnostic codes and invasive procedures, as detailed in Table S1.
We sought to give more robustness to our main analysis by repeating the test in sensitivity analyses, including only people with AD code (331.0) and excluding people with less specific diagnostic codes of senile dementia (290.0, 290.20, 290.21, and 290.3) from the model. Statistical analyses were performed using Stata version 10·1 (Stata, College Station, Texas). Statistical significance was set at P < .05 (2 tailed).
Ethical Aspects
Data confidentiality was maintained at all times in accordance with Spanish legislation. Given the anonymous and mandatory nature of the data set, it was not deemed necessary to obtain informed consent. The study protocol was approved by the ethics committee of the Universidad Rey Juan Carlos.
Results
From 2004 to 2013, we identified a total of 541 858 hospital admissions in Spain in people aged ≥70 years with AD or senile dementia. Over the study period, 26.5% of all admissions occurred in patients with T2D (n = 143 501). We did not observe consistent increases in the rates of admissions over time for any gender (Table 1). Hospitalization rates were significantly higher in people with T2D than in nondiabetic people for all years analyzed (overall, 130.5 vs 91.5 cases per 100 000 people, respectively; Table 2). Taking people without diabetes as the reference category, we observed an overall adjusted incidence rate ratio (IRR, which is the ratio between incidence rates for the populations with T2D and the nondiabetic populations; 95% confidence interval) of hospitalization of 1.41 (1.40-1.42) in people with T2D (men, IRR [95% confidence interval, CI] = 1.32 [1.30-1.33]; women, IRR [95% CI] = 1.46 [1.45-1.48]) using the Poisson regression models.
Rates of Hospital Admissions in People With Alzheimer’s Disease or Senile Dementia (With or Without Type 2 Diabetes and Overall) According to Sex and Age-Groupa.
aAnalysis of the Spanish National Hospital Discharge Database, 2004 to 2013.
bRate values are for hospitalization rates/105 inhabitants.
c P values are for time trends (2013-2004).
dVersus no-T2D, incidence rate ratios [IRRs] (95% confidence interval) for type 2 diabetes are 1.32 (1.30-1.33) in men, 1.46 (1.45-1.48) in women, and 1.41 (1.40-1.42) in both groups combined.
Main Features, Modified CCI, and Coexisting Diagnoses Among Hospitalized Patients With Alzheimer’s Disease or Senile Dementia, According to Type 2 Diabetes Mellitus Statusa.
Abbreviations: n, number of hospital discharges; SD, standard deviation.
aAnalysis of the Spanish National Hospital Discharge Database, 2004 to 2013.
bIncidence rate values are per 105 inhabitants.
cCell values denote n (%), except for age, mean (standard deviation).
d P values are for comparisons by year.
eTo calculate the modified CCI, we have excluded diabetes and dementia.
fThe CCI applies to different disease categories, the scores of which are added to obtain an overall score for each patient.
gWe have grouped patients into 3 categories: low CCI (patients with no previously recorded disease), medium CCI (patients with 1 category), and high CCI (patients with 2 or more disease categories).
Table 2 shows the clinical characteristics in the population with AD or senile dementia according to diabetes status from 2004 to 2013. Two-thirds of the people were women. The proportion of patients older than 84 years was higher among people without T2D, but a higher percentage of people with T2D had a modified CCI ≥ 2. The proportions of people older than 84 years and of people with modified CCI ≥ 2 increased over time (all P values <.001). Malnutrition, pneumonia, and urinary tract infection became more frequently coded in both groups over time (all P values <.001).
Table 3 shows inhospital outcomes and procedures during hospitalization in people with AD or senile dementia according to T2D status. Crude median LOHS and IHM were similar when comparing people with and without T2D. We saw a decrease in both parameters over time both in people with and without T2D (all P values <.001). Time trend analyses (2004-2013) showed significant increases over time in the number of endoscopic gastrostomies, bladder catheterizations, and episodes of mechanical ventilation (all P values <.001).
Hospitalization Outcomes and Diagnostic and Therapeutic Procedures Among Hospitalized Patients With Alzheimer’s Disease or Senile Dementia, According to Type 2 Diabetes Mellitus Statusa.
Abbreviations: CT, computerized tomography; IQR, interquartile range; LOHS, length of hospital stay; MRI, magnetic resonance imaging; PEG, percutaneous endoscopic gastrostomy.
aAnalysis of the Spanish National Hospital Discharge Database, 2004 to 2013.
bCells values denote n (%), except for LOHS, median (IQR).
c P values are for comparisons by year.
dReadmission denotes an admission that took place within the 30-day period after a former hospital discharge.
In Table 4, we can see the results of the multivariate analysis of the factors associated with mortality during hospital admission in people with AD or senile dementia in Spain, 2004 to 2013. Inhospital mortality was significantly higher in men (odds ratio, OR [95% CI] = 1.14 [1.12-1.16]), in older patients (vs 70-74 years old, OR [95% CI] = 1.14 [1.09-1.18] for people aged 75 to 79 years; OR [95% CI] = 1.34 [1.28-1.39] for people aged 80 to 84 years; and OR [95% CI] = 1.79 [1.72-1.86] for people aged ≥85 years), and in those with a higher number of comorbidities (vs no comorbidities, OR [95% CI] = 1.22 [1.19-1.25] for 1 comorbidity and OR [95% CI] = 1.88 [1.83-1.93] for 2 or more comorbidities). Odds ratios were very similar for people with or without T2D.
Factors Associated With Inhospital Mortality Among Hospitalized Patients With Alzheimer’s Disease or Senile Dementiaa.
Abbreviations: CCI, Charlson comorbidity index; 95% CI, 95% confidence interval; OR, odds ratio.
aAnalysis of the Spanish National Hospital Discharge Database, 2004 to 2013.
bTo calculate the modified CCI, we have excluded diabetes and dementia.
cThe CCI applies to different disease categories, the scores of which are added to obtain an overall score for each patient.
dWe have grouped patients into 3 categories: low CCI (patients with no previously recorded disease), medium CCI (patients with 1 category), and high CCI (patients with 2 or more disease categories).
Dehydration, pneumonia, pressure ulcers, parenteral nutrition, readmission, mechanical ventilation, and bladder catheterization were significantly associated with IHM in the population of hospitalized patients with AD or senile dementia, whereas urinary tract infection was associated with survival (Table 4). In our study, T2D was not associated with a higher IHM (OR [95% CI] = 0.99 [0.98-1.01]). Time trend analyses showed a significant decrease in mortality from 2004 to 2013 in patients with AD or senile dementia admitted to hospital (OR [95% CI] = 0.97 [0.96-0.98]) after accounting for T2D status.
In sensitivity analyses, which included people with AD only (ICD-9-CM code 331.0), the results of the multivariate model of the variables associated with IHM did not change substantially (see Table S2), which might suggest that both diagnoses often overlap.
Discussion
Using data from the Spanish National Hospital Database, we found that rates of hospitalization for AD in patients with T2D were significantly higher than in nondiabetic people for all years analyzed. Previous research had underscored the need for improved outpatient care strategies to reduce the impact of comorbidity on unnecessary hospitalization in patients aged ≥65 years with T2D. 20,21 Most formerly published studies had identified infections, dehydration, and malnutrition as the main reasons for hospital admission in patients with dementia. 9 In our study, malnutrition, pneumonia, and urinary tract infection also became more frequently coded over time.
We detected significant increases in the number of endoscopic gastrostomies, bladder catheterizations, and episodes of mechanical ventilation. This might seem paradoxical, in view of the increasing age and number of coexisting medical conditions in hospitalized people with dementia in our country. In a study by Mitchell et al, over 40% of residents of nursing homes underwent at least 1 burdensome intervention (hospitalization, emergency department visit, parenteral therapy, or tube feeding), in the last 3 months of life. 22 In this study, when health-care proxies were aware of the poor prognosis and the expected clinical complications, residents were less likely to undergo these interventions in the final days of life.
We found a ≈40% higher relative incidence of hospitalization in people with T2D as compared to people without diabetes. We can speculate that diabetes increases the risk of clinical events that people with AD or senile dementia are basally more likely to develop, essentially infectious processes. We have previously argued that patients with dementia living in nursing homes might be receiving a lower provision of diabetes-related recommended procedures. 13 Alternatively, we could be merely dealing with some sort of coding bias. Anyway, the design of our study does not let us give a definitive answer to explain this observation. A drawback in our study is that we were not able to identify whether the same person was admitted more than once, except for the 30-day period after admission; thus, the T2D population may be somewhat overrepresented due to a higher number of admissions for any year of the study.
Our time trend analyses showed a significant decrease in mortality from 2004 to 2013 in patients with AD or senile dementia admitted to hospitals in Spain. Others had reported higher IHM rates in the United States for patients with dementia, 1999 to 2009, despite lower hospitalization rates. 23 A more invasive and aggressive clinical management of patients with dementia in more recent years could explain this finding. Although some invasive procedures seem to be associated with mortality in our population, this association possibly obeys to residual confounding and to the fact that more severely ill patients receive a higher number of invasive procedures. Apparently, T2D was not associated with a higher IHM. We have to be cautious about this conclusion, as the data are collected retrospectively and from a registry database. However, speculative plausible explanations include the anti-inflammatory effects of insulin or other treatments, a better tolerance to oxidant stress in some clinical situations, 24 and a lower probability of developing acute lung injury during infections. 25,26
The strength of our findings lies in the large sample size, the 10-year follow-up period covering the population of an entire country, and the standardized methodology, which has been used for research in diabetes and its complications in Spain and elsewhere. 27 Nevertheless, our work is subject to several limitations. Our data source was the CMBD, an administrative database that contains discharge data for hospitalizations in Spain and necessarily relies on the information that physicians include in the discharge report. In previous validation studies of the usefulness of administrative data with research purposes, AD coding is a relatively accurate diagnosis once registered, 28 especially in older people, 29 although it may be somewhat underreported what might have implications for the accurate estimation of its association with risk or protective factors. 30 We decided to include patients with AD and senile dementia codes for the main analyses, even though we could be losing some specificity for the main diagnosis; yet, many cases of AD are actually coded as “unspecific” or “senile” dementia. This fact has sometimes lessened statistical power to draw definite conclusions about this condition in previous research. 23 Moreover, the sensitivity analysis that we did including patients with AD alone basically replicated the findings of the main analyses.
Unfortunately, in Spain, a validation study to assess the rate of unreported diagnosis of diabetes in administrative databases has not been conducted so far. However, Leong et al recently concluded that a commonly used administrative database definition for diabetes had a pooled sensitivity of 82.3% (95% CI: 75.8-87.4) and a specificity of 97.9% (95% CI: 96.5-98.8%), based on the findings of 6 studies with complete data available. 31 Although this definition appears to miss approximately one-fifth of cases with diabetes and wrongly classifies 2.1% of noncases in the population as diabetes cases, it is likely sufficiently sensitive for monitoring prevalence trends in the general population if its accuracy remains reasonably stable over time.
Our findings are also limited by the lack of longitudinal information about diabetes and other variables. The database is also limited by its anonymity (no identifying items, such as clinical history number), which precludes the extraction of some specific pieces of information (ie, people who moved from one hospital to another would appear twice).
In conclusion, Spanish national data show that admission rates were higher in patients with T2D having AD or senile dementia. Inhospital mortality decreased over time in patients with these dementia diagnoses. Diabetes did not independently predict IHM in patients with AD or senile dementia. Given the rapidly increasing prevalence of diabetes and the population aging, our findings emphasize the need to continue improving the clinical management of patients with diabetes having dementia, more specifically by designing strategies to reduce the need for hospital admission and invasive procedures in older patients with advanced dementia.
Footnotes
Authors’ Note
J.M.M.Y. researched data and wrote the manuscript. R.J.G., M.M.B., N.M.R., D.E., J.M.D., and A.L.A. researched data, gave expertise in the field, and reviewed the manuscript. V.H.B. gave statistical support and reviewed the manuscript. All authors meet the criteria for authorship stated in the Uniform Requirements for Manuscripts Submitted to Biomedical Journals. This work has not been published before nor is being considered for publication in another journal. The current research work has been done in accord with the ethical standards of the Committee on Human Experimentation of the Universidad Rey Juan Carlos in Madrid. The sponsors played no role in design, methods, patient recruitment, data collections, analysis, or preparation of the article. The funding sources had no involvement at any step of the research done or in the process of writing this manuscript.
Acknowledgments
The authors wish to thank the Spanish Ministry of Health, Social Services and Equality for providing the records of minimum basic data set (MBDS).
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Dr De Miguel-Yanes reports speaker fees from Novo-Nordisk, Boehringer-Lilly, Astra-Zeneca, Sanofi-Aventis, and Ferrer, outside the submitted work. This study is part of research funded by the FIS (Fondo de Investigaciones Sanitarias-Health Research Fund, grant no. PI16/00564, Instituto de Salud Carlos III) cofinanced by the European Union through the Fondo Europeo de Desarrollo Regional (FEDER, “Una manera de hacer Europa”).
Supplemental Material
Supplementary material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
