This study demonstrates that hospitalization is a frequent outcome in persons with Alzheimer’s disease, even at relatively early stages of disease. In our sample of community-dwelling persons with (on average) mild AD and low comorbidity, two-thirds were hospitalized at least once during the follow-up period, and nearly half were hospitalized 2 or more times. Due to longer duration of follow-up (3.0 years on average), these cumulative risks are higher than those described in previous studies of community-dwelling samples.11
Our results reinforce previous studies reporting on the high rates of hospitalization in AD patients more generally, which are at least 3 times higher than non-affected, age-matched persons.13, 26, 27
This study has face validity in that three of our independent risk factors for hospitalization (age, comorbidity, and prior hospitalization) have been identified in other studies of the Alzheimer’s and older primary care populations.15, 26
. Our study expands the literature by identifying two additional risk factors for hospitalization which are routinely assessed in Alzheimer’s Disease patients (male sex and short duration of symptoms) The short duration of symptoms before diagnosis indicates that patients on a rapid or accelerated trajectory of decline before presenting for dementia evaluation may be more likely to be hospitalized. Interestingly, in this community-dwelling population severity of dementia and other dementia-related variables including cognitive score, dementia severity rating, and clinical course were not significant risk factors for hospitalization. While some studies have found that more severe dementia was predictive of hospitalization, the severity index is not always collected systematically.11–13
For example, dementia severity measured at the time of hospitalization will be higher because of the physiologic decompensation and possibly superimposed delirium28
which brought the patient to the hospital. By measuring severity at ADRC visits prior to hospitalization, we are capturing a picture of the baseline dementia severity, rather than acute dementia severity.
Taken together, the risk factors identified help to identify a high risk group. For instance, patients with 2 or more risk factors have a >70% cumulative risk of hospitalization over approximately three years, and would be an important group to target for future intervention strategies. While validation in other samples would be needed, our results describing the risk factors for hospitalization yield important and clinically relevant information. While speculative, it is possible that early stage AD patients are more likely to be hospitalized than their age-matched peers because of increasing problems related to their cognitive deficits, even at the early stages, such as judgment errors resulting in falls, medication errors or noncompliance, decreases in thirst or appetite resulting in dehydration or malnutrition, and the like. Even in the early stages, it is likely that subtle cognitive problems may have adverse impacts on health and medical care.
Understanding potential factors leading to hospitalization in the AD population are important for quality of life, cost-savings, and development of interventions to prevent hospitalization. Many of the identified conditions are potentially preventable with simple interventions in the home and hospital, and with discharge teaching/homecare follow-up. The five leading reasons for hospitalization in this cohort included falls, ischemic heart disease, gastrointestinal disease, pneumonia, and delirium which are consistent with recent work.15, 29
Falls may be effectively prevented in community-dwelling elderly through strategies such as exercise programs, balance training, reduction of high-risk medications, vision adaptations, and home safety enhancements.30
Prevention strategies for ischemic heart disease include blood pressure control, cholesterol reduction, smoking cessation, diet modification, and increased physical activity.31–33
These interventions would also reduce cerebrovascular events, the sixth most common reason for hospitalization. For gastrointestinal bleeding and peptic ulcer disease, leading gastrointestinal diagnoses in our sample, avoidance of non-steroidal drugs, use of antacids or gastroprotective drugs, and eradication of H pylori
, may be effective preventive strategies.34
Strategies for prevention of pneumonia, including improved swallowing techniques, aspiration precautions, avoidance of sedative medications, consistent use of influenza and pneumococcal vaccination, and prompt use of appropriate anti-viral treatment during flu season, may help to reduce these admissions.35
Finally, well-established strategies for delirium prevention, including avoidance of psychoactive medications, enhancing nutrition and hydration, preventing infection, addressing electrolyte or metabolic derangements, and ensuring oxygenation, would be effective approaches to reduce hospitalization.36
It is important to note that many of these evidence-based strategies have not been tested specifically for their preventive efficacy in persons with Alzheimer’s disease.
This study has noteworthy strengths. It represents a unique large-scale epidemiologic study examining the frequency and risk factors for hospitalization in a community-dwelling AD population. High quality data from the MADRC provided a large, well-characterized clinical cohort of 827 patients, with prospective gold-standard neurologist diagnoses of AD according to clinical criteria and longitudinal follow-up for a median of 3 years. Furthermore, the careful combination of MADRC and MedPAR data allowed near-complete examination of all hospitalization data across several hospitals and states.
The results of this study are limited because we were unable to fully assess acute indications for hospitalization such as delirium, medication, and social factors including caregiver support from the MADRC assessments and Medicare data. Additionally, for the initial year of the study (1991), we did not have prior hospitalization records. A sensitivity analysis using multiple imputation of these records did not significantly change the results. Finally, this study would have benefited from further assessment at the index MADRC visit of the relationship between speed of cognitive decline and recent delirium.
The generalizability of these results are limited by several factors listed below. First, the exclusion of patients with <3 MADRC visits may have limited our generalizability to patients earlier in the disease and those who were more likely to follow-up. Second, the minority representation in this sample is relatively low (6%). While we have verified that the MADRC sample is comparable to the National Institute on Aging/National Alzheimer’s Coordinating Center (NACC) sample of 85,460 AD patients in terms of age, gender, educational level, and dementia onset, the minority representation is lower in the MADRC than in the overall NACC sample (6% vs.18%). Third, the data were drawn from a single site which may limit generalizability to patients who have access to an academic medical center or ADRC. However, the primary care and thus, most of the decisions regarding hospitalization of the AD patients were made by the primary care physicians. Finally, it is important to stress that all of these factors influence only external validity (generalizability), and the internal validity of our study findings remains uncompromised.
There are important implications of our work. First, the study allows us to identify AD patients at high risk for hospitalization, who would be important to target for future preventative efforts. Second, the study lays the groundwork for future intervention studies to prevent hospitalization in this vulnerable group. Finally, the outcomes of hospitalization in AD patients warrant further examination in future studies. Documenting the prevalence and etiologies of hospitalization in AD patients represents a crucial first step in this future research imperative. Given the frequency of hospitalization in the currently 5 million patients with AD, an intervention to prevent or shorten hospitalization would potentially save Medicare billions of dollars per year.5, 37