https://orcid.org/0000-0002-3309-618X
Figures
Abstract
Alzheimer’s disease (AD) imposes a major burden on affected individuals, their caregivers and health-care systems alike. Though quite many risk factors for disease progression have been identified, there is a lack of prospective studies investigating the interplay and predictive value of a wide variety of patient variables associated with cognitive deterioration (defined as key feature of AD progression). Study participants were patients with probable and possible AD, that were assessed at four time points over a period of two years (T1-T4). The main results were threefold: (i) over time, significant changes were observed regarding patients’ cognitive functioning, activities of daily living and caregiver load (but not depression, pain, neuropsychiatric symptoms); (ii) intercorrelations between caregiver load and patients’ cognitive and functional variables were high, correlation patterns remaining rather stable across time; (iii) cognitive functioning at T4 was best predicted by patients’ age, sex, atrial fibrillation and activities of daily living at T1; and (iv) across all four assessment points, cognitive functioning was best predicted by time (i.e., disease duration), age, sex, activities of daily living and depression. Overall, even in early stages of AD and during a short two-year period, functional changes were significant and tightly intertwined with caregiver load, thus stressing the need to consider caregiver load when diagnosing and treating patients with AD. A novel and clinically relevant finding is that even in early stages of AD, cognitive deterioration was best predicted by a combination of patients’ demographic, somatic and functional variables.
Citation: Kaufmann L, Gruenbaum T, Janssen R, Weiss EM, Benke T, Dal-Bianco P, et al. (2024) Predictive value of somatic and functional variables for cognitive deterioration for early-stage patients with Alzheimer’s Disease: Evidence from a prospective registry on dementia. PLoS ONE 19(8): e0307111. https://doi.org/10.1371/journal.pone.0307111
Editor: Thiago P. Fernandes, Federal University of Paraiba, BRAZIL
Received: January 31, 2024; Accepted: June 30, 2024; Published: August 14, 2024
Copyright: © 2024 Kaufmann et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: The underlying data can be found at the OSF repository. The DOI is as follows: 10.17605/OSF.IO/R7HSF.
Funding: The work of RJ was supported by a grant from the French National Research Agency (ANR-22-CE28-0020; https://anr.fr/Project-ANR-22-CE28-0020).
Competing interests: The authors have declared that no competing interests exist.
Introduction
Alzheimer´s disease (AD) is the most common form of dementia, affecting 50 to 70% of patients diagnosed with dementia [1]. Importantly, the high–and with increasing life expectancies still rising—incidence rates of AD [2,3] are associated with high public health costs as affected individuals require comprehensive interdisciplinary attention (including medical, therapeutical, palliative care).
Upon acknowledging the societal impact of AD (including the associated challenges and costs for medical diagnosis and treatment), it is not surprising that there is a vast literature on potential risk factors for AD (other than old age which previously has been identified a primary risk factor for AD) [3,4].
Somatic and functional risk factors
Concerning somatic risk factors for cognitive impairment and incident dementia, findings of recent meta-analyses reveal that patients suffering from coronary heart diseases are at increased risk to develop cognitive impairment and dementia [5,6]. Moreover, there is accumulating evidence that hypertension, atrial fibrillation, diabetes mellitus and hypercholesterolemia are important risk factors for cognitive decline, eventually leading to AD [7; see also 8 and 9 for comprehensive and critical overviews of potentially modifiable risk factors for dementia]. Notably, successful treatment of vascular risk factors such as lowering high blood pressure has been reported to be significantly related to a lower risk of cognitive decline and incident dementia [10,11]. Because previous findings were targeted at examining the effects of systolic blood pressure interventions on cognitive deterioration in dementia [12], we decided to consider both systolic and diastolic blood pressure in our analyses. Importantly, as old age frequently is associated with several of the above-mentioned somatic risk factors, the negative effects of each of these disorders may overlap and synergize, thus heightening the risk of cognitive impairment and AD [7].
Beyond somatic risk factors, also functional non-cognitive variables have been identified as potential risk factors or factors thought to facilitate disease progression. Above all, depression (beyond age) has been identified as a major risk factor for the development of AD [13] and moreover, depressive symptoms and depression severity were found to facilitate and even predict the conversion from mild cognitive impairment to AD [14,15]. Recent findings of a large-scale prospective cohort-study disclosed that depression is associated with a 51% higher risk of dementia [16]. Furthermore, depression is a major burden for elderly individuals with dementia that suffer from chronic pain [17–19]. Likewise, chronic pain in elderly–especially in association with dementia-related cognitive changes–imposes a high burden on affected patients and their family caregivers alike [20]. Even though chronic pain is a frequent sequalae of age-associated accumulating somatic complaints (including those related to the movement apparatus) the management and treatment of pain in dementia has been hardly studied [17,18]. Although some evidence suggests that pain in the elderly is linked to neuropsychiatric symptoms, the evidence for this relation is controversially discussed [17,20]. However, the presence of neuropsychiatric symptoms–particularly those in the psychotic cluster–has been identified as a reliable indicator of cognitive decline (measured by the Mini-Mental-State-Examination/MMSE) in individuals with dementia [21]. Finally, there is accumulating evidence suggesting that patients’ increasing functional dependency (as reflected in impaired activities of daily living/ADLs) exerts a negative impact on caregiver’s physical and psychological wellbeing, thus resulting in growing caregiver load [22,23]. Not surprisingly, accumulating evidence suggests that patients’ increasing functional impairments (e.g., reduced ADLs, chronic pain and neuropsychiatric behaviors) exert direct and indirect effects on caregiver’s wellbeing [23,24].
Finally, longitudinal large-scale studies found that up to half of the informal caregivers (e.g., family members, neighbors, friends) of patients diagnosed with AD report clinically significant levels of challenges related to the caregiving situation [22,25], proportions even increasing with AD progression. Interestingly, both caregiver and patient characteristics might be predictors of caregiver load. Converging evidence suggests that the most reliable indicators of patient outcome are the severity of cognitive impairment, neuropsychiatric behavioral symptoms and functional dependency [22,24,25]. Moreover, risk factors for developing caregiver load included higher age, female sex and living together with the patient [24,25]. Clearly, the presence of neuropsychiatric symptoms (that can be differentiated into behavioral, emotional, and psychotic behaviors [21]) and/or higher functional dependency (e.g., increasing need for support to manage daily affairs) is unequivocally associated with an increasing impact on caregivers’ physical and mental health [22,24–26]. Hence, the examination of the interplay between caregiver load and patient-related measures—especially during early stages of AD—is of utmost interest.
Study aims
The focus of the present prospective study is on demographic (age, sex, and education) and somatic potential risk factors (atrial fibrillation, coronary heart disease, hypercholesterolemia, hypertension, diabetes, diastolic and systolic blood pressure) as well as non-cognitive functional variables frequently related to AD (functional dependency as indicated by decreasing activities of daily living/ADLs, depression, pain, and neuropsychiatric symptoms). Moreover, upon acknowledging findings suggesting a correlation between caregiver load and AD patients’ health status [22,24,26], also subjective caregiver load was introduced into our longitudinal analyses.
In the present study, we report data from the Prospective Registry on Dementia (PRODEM) in Austria. The PRODEM is a prospective cohort study that followed patients with AD and related dementias over a period of two years across four assessment points (i.e., baseline, six months after baseline as well as one and two years after baseline). The main aims of the present longitudinal study were as follows: (i) to investigate early-stage disease progression of AD (indexed by patients’ functional -including cognitive- variables), (ii) to examine correlation patterns and correlation strengths between patients’ functional -including cognitive- variables and caregiver load across time, and finally (iii) to identify the predictive value of patients’ demographic, somatic and functional risk variables at baseline on patients’ cognitive functioning at the last follow-up examination as well as across all four assessment points (see Fig 1).
Abbreviations: CERAD = Consortium to Establish a Registry for Alzheimer’s Disease; DAD = Disability Assessment for Dementia (indexing activities of daily living/ADLs); GDS = Geriatric Depression Scale; Pain = Pain scale focusing on pain during activity; NPI = Neuropsychiatric Inventory; ZBI = Zarit Burden Inventory. Notes: Somatic risk factors were recorded at baseline only (T1), while cognitive functioning (indexed by CERAD), functional non-cognitive risk factors and caregiver load were examined at all four assessment points (i.e., from T1 to the last follow-up two years after baseline/T4).
Materials and methods
Ethics statement
Ethical approval was obtained from the ethics committees of the following participating centers: Medical University of Graz, Medical University of Innsbruck, Medical University of Vienna, Konventhospital Barmherzige Brüder Linz as well as the Provinces of Upper Austria, Lower Austria and Carinthia. Please note that in Austria, it is sufficient to obtain ethical approval from the leading study center. The leading study center in the present study was the Medical University of Graz, thus, ethical approval was obtained from the ethics committee of the Medical University of Graz (approval number 19–135 ex 07/08) and accepted by the ethics committees of the other participating study centers.
Study participants
All participating patients and their caregivers were recruited via the multi-centric cohort study “Prospective Dementia Registry Austria (PRODEM-Austria)” of the Austrian Alzheimer Society. Data collection for the present study started on January 23, 2009 and ended April 21, 2017. Written informed consent was obtained from all study participants. All participating patients and caregivers underwent a comprehensive clinical examination at baseline (T1) and were invited for regular clinical routine follow-up examinations six month after baseline (T2) as well as one and two years after baseline (T3 and T4, respectively).
Inclusion criteria.
Inclusion criteria were as follows: living still at home, availability of a family caregiver who could provide information on the patient’s neuropsychiatric symptoms and health condition. Excluded were patients with reduced capacity to consent and/or who were unable to sign an informed consent or if co-morbidities were likely to preclude study termination. Importantly, reduced capacity to consent was an exclusion criterion at all time points (i.e., T1 to T4). In our sample, no single patient had to be excluded at the follow-up visits because of severe cognitive deterioration yielding reduced capacity to consent and/or difficulties to understand the medical and neuropsychological procedures associated with the follow-up visits. The participating study sites were federal or provincial hospitals specialized in neurology or psychiatry and located in six of the nine provinces in Austria. Upon applying the latter criteria, the study sample consisted of 578 patients diagnosed with possible or probable AD according to NINCDS-ADRDA criteria [27,28]. Due to missing data (i.e., not all patients completed all tests at the four assessment points) we decided to include only patients into the analyses for whom we had data from the cognitive outcome measure (CERAD) and at least one non-cognitive functional variables at each assessment point (i.e., T1 to T4; see the section”Functional non-cognitive measures” below). Hence, at baseline, the initial study sample comprised N = 500 patients.
Data acquisition.
First, all patients underwent clinical examinations that involved the acquisition of demographic data (Table 1) and results of functional (including cognitive) tests. Moreover, patients’ medical records were used to obtain information regarding somatic risk factors (Table 2). As outlined in the introduction, the variable selection was based on the literature and thus, is selective rather than comprehensive. In addition, family caregivers were asked to assess patients’ functional dependency (i.e., activities of daily living/ADLs) and potential behavioral changes (i.e., neuropsychiatric symptoms, known to be frequently associated with dementia) and finally, were asked to report on their caregiver load (see Table 1 for caregivers’ demographic data).
Reported are the number of cases with a specific somatic diagnosis (n reported/n not reported).
Drop-out analysis.
Across the four study time points, the number of participating patients decreased from n = 500 at baseline (T1) to n = 169 at T4 (i.e., two-year follow-up; Table 1). Though this drop-out rate is rather high, it is according to expectation as in clinical longitudinal studies, patients are frequently lost due to translocation, change of medical treatment/hospital, disease progression hampering study compliance, etc. [29–31].
Importantly, the drop-out analysis disclosed no significant differences between patients who were lost to follow-up and those who remained in the longitudinal study regarding demographic variables (i.e., age, sex, and education), a cognitive screening test (i.e., MMSE) and all but two somatic variables (i.e., atrial fibrillation, coronary heart disease, hypertension, diabetes, and systolic blood pressure). The only significant group differences emerged as regards hypercholesterolemia and diastolic blood pressure (Table 3). Moreover, significant group differences between patients who dropped out and those who remained in the study were observed regarding caregiver load (indexed by the Zarit Burden Inventory/ZBI).
Cognitive and non-cognitive functional measures (tests and questionnaires)
The present longitudinal study comprised four assessment times (i.e., baseline followed by three follow-up examinations). Please note that for all measures (i.e., the cognitive outcome measure CERAD and the non-cognitive functional variables) solely the total scores were entered into the data analyses.
Overall cognitive functioning.
Overall, cognitive functioning was assessed by asking all patients to complete the CERAD neuropsychological test battery (Consortium to Establish a Registry for Alzheimer’s Disease [32]) which includes subtests measuring semantic verbal fluency, a confrontation naming test (i.e., modified Boston Naming Test), a verbal memory test (tapping word list learning, word list immediate and delayed recall, word list recognition), a constructional praxis test (copying figures), a figural memory test (delayed recall) as well as the MMSE [33]. Out of these measures, we calculated a CERAD total score (age and education corrected z-scores [34]). The CERAD total score was used as cognitive outcome measure and served as a proxy for disease progression. Overall, higher z-scores indicate better cognitive performance (and vice versa).
Activities of daily living.
Activities of daily living (ADLs) were assessed by using the scale Disability Assessment for Dementia (DAD) [35,36]. The DAD requires caregivers to report on patients’ basic and instrumental activities of daily living as well as leisure activities over the last two weeks. Overall, the DAD consists of 40 items that assess whether a specific ADL is still preserved or not (YES vs NO answer, respectively). A third answer category is reserved for items that are not applicable (NA). Please note that NA answers are not considered and thus, reduce the maximum number of items. The DAD total score is provided as a percentage (number of YES items x 100 divided by the maximum number of items). In the present study, the total DAD score was used to measure the patients’ ADLs, with lower scores indexing higher levels of functional dependency (and higher scores indicating higher levels of functional independency).
Depressive symptoms.
Depressive symptoms were examined by using the short version of the Geriatric Depression Scale (GDS-15) [37,38]. The GDS short version consists of 15 items that require a YES or NO answer (max. score 15). Higher scores indicate a more pronounced depressive symptomatology.
Pain.
Pain was assessed using a structured pain questionnaire specifically developed for geriatric patients (developed and registered by the German Pain Society) [39]. The original pain interview consists of questions requiring the patient to provide information regarding pain localization (including number of body parts affected by pain), pain intensity and functional impairments caused by pain (i.e., hampering patients’ ADLs). Based on the literature suggesting that pain during activity is a frequent complaint among the elderly, both with and without dementia [40,41], we chose to focus our analyses on questions tapping pain during activity. Overall, pain during activity was assessed by four items (taken from Basler [39]). Participants were required to answer YES if they were still able to perform the specific activity, or alternatively, to answer NO if they were unable to perform the activity due to pain or due to other reasons. The total pain score was derived by adding the NO answers due to pain only, thus yielding a total score of 4, with higher scores indicating more pain.
Neuropsychiatric symptoms.
Neuropsychiatric symptoms were evaluated using the Neuropsychiatric Inventory (NPI) [42]. The NPI assesses 12 subdomains of behavioral functioning (i.e., delusions, hallucinations, agitation/aggression, depression/dysphoria, anxiety, irritability/lability, euphoria, apathy, disinhibition, aberrant motor behavior, night-time behavioral disturbances, appetite and eating abnormalities). In the present study, we used the total NPI score, that is derived by multiplying symptom frequency (scored on a 4-point scale ranging from 0 (never) to 4 (at least once per day)) with symptom severity (scored on the 3-point scale mild, moderate and severe) for each of the 12 subdomains separately. Hence, the maximum raw score is 144 (12 symptoms x 4 x 3), with higher scores indicating more neuropsychiatric symptoms.
Caregiver load.
Caregiver’s subjective feelings of challenges related with the caregiving situation were measured by using the German-language version of the Zarit Burden Inventory (ZBI) [43,44]. The ZBI consists of 22 items related to patients’ behavioral or functional impairments and/or the care situation. Items were answered on a 5-point Likert scale (ranging from 0 to 4). Thus, the total score ranged from 0 to 88, with higher scores indicating stronger feelings of burden.
Statistical methods
All analyses were conducted by using the statistic software R (version 4.3.0) [45]. For all analyses, results were considered significant if p < 0.05. At each timepoint, we included all patients who had a CERAD score and, moreover, a valid measurement of at least one additional functional test (i.e., DAD, GDS, PAIN, NPI).
Measures.
For all functional tests (i.e., cognitive outcome measure CERAD as well as all non-cognitive functional tests), we used the total raw scores as dependent variables. The dependent variables of the somatic patient variables were–alike the functional patient variables—either interval scaled (i.e., diastolic and systolic blood pressure) or dichotomous (i.e., yes/no: atrial fibrillation, coronary heart disease, hypercholesterolemia, hypertension, diabetes). Likewise, demographic variables were interval scaled (i.e., age, education in years) or dichotomous (sex).
Statistical analyses.
Beyond reporting on the descriptive results of the functional variables (including the CERAD total score used as a cognitive outcome measure, see data reported in S1 Fig in S1 File), we calculated linear mixed models—using the lme4 package (version 1.1.34 [46]; time being included as a fixed effect, while patient-ID was entered as a random effect) aiming at investigating whether functional changes over time were significant (Table 4).
Values in squared brackets indicating 95% confidence intervals.
In a next step, Pearson correlation analyses were conducted (by using the Hmisc package, version 5.0.1 [47]) to examine whether correlation strengths and patterns between the functional (cognitive and non-cognitive) variables changed across the four assessment points (Table 5).
Finally, linear regression analyses were performed to identify potential causal relationships between the cognitive outcome measure and demographic, somatic and functional patient variables (see Tables 6 and 7). As depicted in Table 6, a bidirectional stepwise linear regression analysis was conducted (using the caret package, version 6.0.94 [48]) to examine whether the dependent variable (i.e., cognitive outcome measure as indexed by the CERAD total score at T4) can be predicted by any of the predefined independent variables at T1 (i.e., patients’ demographic, somatic and functional risk variables). Predictors were sequentially added and removed to find the set of predictors best suited for variance explanation. Moreover, 5-fold cross-validation was used to ensure that the result from the stepwise regression would be unbiased and reliable. Finally, the regression model with the best fit to our data (i.e., the lowest ‘root mean square error of approximation’/RMSEA) was then used for the following analyses. The final model was checked for collinearity and potential interactions between these predictors. Importantly, our findings from stepwise regression analysis seem to be robust and not biased by any potential interaction effects between the predictor variables.
In addition (see Table 7), by using linear mixed models (treating time as fixed effect and patient-ID as random effect) we aimed at investigating whether, across the four assessment points, functional risk factors were predictive of cognitive functioning (entered as dependent variable and indexed by the CERAD total score at each time point).
To counteract the problem of multiple comparisons (i.e., type 1 error), we applied Bonferroni corrections to the drop-out and the correlation analyses.
Upon interpreting whether statistically significant results were also clinically meaningful, we used the benchmarks proposed by Cohen [49], suggesting that Pearson’s correlation coefficients r between 0.1 and 0.3 indicate small effect sizes, those between 0.3 and 0.5 medium and those ≥ 0.5 large effect sizes (which correspond to Cohen’s d of 0.2, 0.5 and 0.8, respectively). Likewise, upon interpreting the regression results (i.e., the coefficients of determination R2 indicating the percentage of variance of the dependent variable explained by the independent variables), we follow Cohen’s [49] suggestion proposing that R2 of 0.02 reflect small, 0.13 medium and 0.26 strong effect sizes.
Results
In the following, we will elaborate our findings along the three main research questions as outlined in the section “Study aims”.
Disease progression in the early stages of AD
To examine whether cognitive and non-cognitive functional changes across time (i.e., from the baseline assessment at T1 to the last follow-up two years later at T4) were significant and clinically meaningful, we estimated linear mixed models for each functional variable (Table 4). Interestingly, significant changes over time were observed regarding CERAD (overall cognitive functioning), thus reflecting poorer cognitive abilities with increasing disease duration (for the descriptive results, see data in S1 Fig in S1 File of the Supporting Information). Moreover, DAD (tapping ADLs) and ZBI (caregiver load) changed significantly across time. As depicted in the data in S1 Fig in S1 File, DAD scores decreased significantly over time, indicating higher functional dependency, and higher ZBI scores reflected higher caregiver load. In contrast, ratings of GDS (depressive symptoms), PAIN (tapping pain during activity), and NPI (neuropsychiatric symptoms) did not reach significance.
Correlation between patient-related variables and caregiver load over time
The strongest and highly significant correlations were observed between caregiver load (ZBI) and patients’ ADLs (DAD), neuropsychiatric symptoms (NPI) and cognitive functioning (CERAD). Although correlation patterns remained rather stable over time, correlation strengths changed somewhat across the four assessment points (Table 5, also see data in S2A-S2E Fig in S1 File). Higher caregiver load was associated with poorer ADLs (correlation coefficients > -0.50 at all assessment points), more neuropsychiatric symptoms (correlation coefficients ranging from 0.42 to 0.45), and lower cognitive functioning (correlation coefficients ranging from -0.14 to -0.35). Furthermore, as depicted in Table 5, caregiver load was significantly and positively correlated with depressive symptoms (GDS) at three out of the four time points (i.e., T1, T3, and T4). Another highly significant correlation emerged between cognitive functioning (CERAD) and ADLs. Most notably, across time the latter correlation became increasingly stronger (correlation coefficients ranging from 0.23 to 0.46; positive correlation coefficients indicating that lower CERAD scores being associated with poorer ADLs as indexed by lower DAD scores). In addition, CERAD scores were significantly correlated with patients’ depressive symptoms (GDS) at T2 and T3 (correlations being non-significant at T1 and T4), reflecting that lower CERAD scores went along with higher GDS scores (i.e., more depressive symptoms). Furthermore, CERAD scores were significantly correlated with neuropsychiatric symptoms (NPI) at T1 and T4 (correlations remaining non-significant at T2 and T3), indicating that lower CERAD scores were associated with higher NPI scores (i.e., more neuropsychiatric symptoms). Furthermore, at three out of four time points (i.e., from T1 to T3), neuropsychiatric symptoms (NPI) were positively and significantly correlated with depressive symptoms (GDS). Finally, there was a correlation between pain and caregiver load at T1 (r = 0.15) but the correlation did not reach significance at the follow-up visits (see Table 5 for a full overview of all correlations).
Predictive variables for cognitive deterioration
First, upon employing stepwise regression analysis, we sought to examine whether and which patient variables at T1 (i.e., demographic, somatic and functional non-cognitive variables) might be predictive of cognitive deterioration (i.e., cognitive functioning as indexed by CERAD score at T4). Second, by employing a linear mixed model, we sought to identify patient variables (i.e., demographic and non-cognitive functional risk factors) that might be significant predictors of cognitive functioning across the four follow-up visits (i.e., from T1 to T4). Please note that somatic patient variables were not entered into the latter analyses, as somatic variables were collected at T1 only.
Results of the stepwise regression analysis disclosed that patients’ cognitive functioning at T4 was significantly predicted by patients’ age and sex as well as atrial fibrillation and ADLs (i.e., functional dependency as indexed by DAD) at baseline (Table 6, see also data in S3 and S4A-S4D Figs in S1 File). The overall model fit disclosed an adjusted R2 = 0.137 (corresponding to a medium effect size according to Cohen [49]).
As depicted in Table 7, results of the linear mixed model disclosed a good overall fit (marginal R2 = 0.157, corresponding to a medium effect size according to Cohen [47]) and disclosed that cognitive functioning is best predicted by time (i.e., disease duration), age, sex, ADLs (i.e., functional dependency as indexed by DAD) and depressive symptomatology (GDS). More specifically, increasing time (being a proxy for disease duration/progression), high age, female sex, higher functional dependency and more severe depressive symptomatology were found to be associated with lower cognitive functioning.
Discussion
In this study, we examined the impact of various somatic and non-cognitive functional patient-related variables on the progression of early Alzheimer’s disease (AD) as measured by a cognitive outcome measure (i.e., CERAD total score). Furthermore, we investigated the relationship between these patient-related variables and the level of caregiver load. Our primary findings were as follows: First, significant changes in functional capacity were observed over time in relation to patients’ cognitive functioning (i.e., CERAD total score [34]), patients’ ADLs and caregiver load. However, no significant changes were observed in the remaining functional variables (such as depressive symptoms, pain, and neuropsychiatric symptoms) between the baseline (T1) and the final follow-up visit two years after baseline (T4). Second, the correlation analyses revealed a strong and consistent relationship between caregiver load and patients’ cognitive and non-cognitive functional variables over the two-year study period. Notably, caregiver load was found to be significantly and positively correlated with patients’ ADLs, neuropsychiatric symptoms, cognitive functioning (CERAD), and depressive symptoms) across all four assessment points. Third, regression analyses showed that among all patient variables collected at T1, only age, sex, atrial fibrillation, and ADLs were significant predictors of cognitive functioning at T4 (measured by the CERAD total score). Additionally, across all four timepoints, patients’ cognitive functioning was best predicted by time, patients’ age, sex, ADLs, and depressive symptoms. In the following, we will discuss these findings in greater detail and consider their implications for our understanding of AD and its progression.
Disease progression during early stages of AD
Regarding cognitive decline (as reflected in changes of CERAD scores from T1 to T4), our findings contribute to the existing body of literature [1,3,4] by revealing the significant deterioration of cognitive abilities even in the early stages of AD. This decline occurs over a relatively short period of two years. Additionally, our findings disclosed notable changes in patients’ ADLs, which reflect their increasing functional dependency as measured by the DAD. Furthermore, significant changes were observed in subjective feelings of caregiver load, as indicated by the ZBI, throughout the two-year follow-up. While not entirely unexpected, our findings disclose that even in the early stages of the disease, significant decreases in ADLs are associated with significant increases in caregiver load. Therefore, our findings, in conjunction with the correlation analyses discussed in the subsequent section, further validate the close relationship between patients’ non-cognitive functional dependency and caregiver load [22,24–26,50]. Furthermore, the seemingly insubstantial, yet quantifiable augmentation of neuropsychiatric symptoms (indexed by the NPI) indicates that increasing disease duration is correlated with an amplified manifestation of neuropsychiatric symptoms [21,51]. Our findings partially resemble prior research indicating significant changes over time in NPI ratings related to one of the three NPI subsyndromes (specifically, the psychotic subsyndrome), while no significant alterations were observed over time in the behavioral and emotional NPI subsyndromes [21]. However, the total NPI score was not provided in the aforementioned study, thereby rendering a direct comparison unfeasible. Conversely, and somewhat unexpectedly, patients’ self-assessments of pain levels remained relatively low and displayed negligible fluctuations over time. The reasons for this phenomenon remain ambiguous and may range from diminished pain perception to improved pain management techniques [18–20].
Correlation between patient-related variables and caregiver load
As depicted in Table 5, we observed the strongest and highly significant correlations between caregiver load and patients’ ADLs, neuropsychiatric symptoms and cognitive functioning. Higher caregiver load was associated with poorer ADLs, more neuropsychiatric symptoms, and lower cognitive functioning. Over a two-year period (i.e., at T1, T3 and T4), caregiver load was consistently correlated with depressive symptoms. Strong interrelations were also found among patients’ cognitive and non-cognitive functional variables. Poorer ADLs were consistently correlated with lower cognitive functioning and more neuropsychiatric symptoms. Thus, our findings highlight the significant and stable interplay between caregiver load and patients’ functional risk factors during early stages of AD. This emphasizes the importance of systematic evaluations of cognitive and non-cognitive functional impairments in clinical practice. In addition, our results align with previous studies indicating a strong link between caregiver load and patients’ behavioral symptoms, especially disruptive behaviors as measured by the NPI [26], as well as non-cognitive functional decline, including impaired ADLs [22,25]. However, a unique feature of our study is its prospective design, covering a two-year follow-up period, and examining somatic and functional non-cognitive risk factors as potential predictor variables for cognitive decline. We specifically focus on the early stages of AD, conducting a combined and longitudinal investigation of caregiver load and fundamental non-cognitive patient functions that affect AD progression, such as ADLs, depressive symptoms, neuropsychiatric symptoms, and pain related to activity.
Moreover, our results support the idea that a systematic informant-based assessment of functional deficits serves as a valuable clinical indicator of early cognitive deterioration [52–55]. The recognition of impaired ADLs as a significant phenotypic marker for distinguishing minor from major neurocognitive disorders (DSM-V [56]) underscores the importance of conducting regular ADL assessments in clinical settings.
Predictive variables for cognitive deterioration
Our primary objectives were to identify predictors of cognitive deterioration (CERAD at T4) by using stepwise regression at T1. Additionally, a linear mixed model was employed to discern predictors of cognitive functioning across all four assessment points over time. The stepwise regression analysis yielded a well-fitted model (adjusted R2 = 0.137, indicating a medium effect size per Cohen’s benchmarks [49]; see Table 6). Approximately 14% of the variance in patients’ cognitive functioning (at T4) could be explained by four predictors identified two years earlier (at T1): higher age, female sex, the presence of atrial fibrillation, and poorer ADLs were all associated with poorer cognitive functioning. Furthermore, the linear mixed model results indicated that patients’ cognitive functioning over time (T1 to T4) is best predicted by time, age, sex, ADLs, and GDS (Table 7). The model showed a good fit (marginal R2 = 0.157, a medium effect size per Cohen [49]), suggesting that approximately 16% of the variance in patients’ cognitive functioning can be explained by increasing time (a proxy for disease progression), higher age, female sex, poorer ADLs (DAD), and more severe depressive symptoms (GDS). Notably, this analysis did not include somatic risk factors that were collected at T1 only (see Fig 2 for a schematic representation of the results of the two regression analyses). Overall, our model fits (suggesting medium effect sizes according to Cohen [49]) are good, especially upon acknowledging that our findings are complex and influenced by many factors, some of which we were able to assess, while we did not examine other factors that previously have been reported to influence cognitive deterioration such as lifestyle factors (physical activity, nicotine and alcohol consumption, cognitive (in)activity etc. [8,9]. Hence, although we entered quite many potential predictor variables to our stepwise regression model initially, our model might not be complete as we were not able to include all potentially relevant predictor factors for cognitive deterioration. Further potential reasons for our–at first glance—rather modest model fit are the following: (i) our prospective study aimed at investigating disease progression in early-stage dementia over a relatively short period of two years (which is very brief upon considering that dementia per definition is a slowly progressive neurodegenerative disease [1–4,57]); (ii) the high drop-out rate clearly reducing our model fit; and (iii) some of the measures and tests used were observational and/or based on subjective patient/caregiver reports and thus, possibly less sensitive than objective assessment tools. However, variance explanations of 14% and 16% (see Tables 6 and 7, respectively) are quite remarkable upon acknowledging that the functional changes were observed in early stages of a neurodegenerative disease and within a short time span of two years.
(A) Upper panel showing the findings of the stepwise regression investigating the predictive value of all patient variables at T1 (demographic, somatic and functional) on cognitive functioning at T4. (B) Lower panel showing the findings of the linear mixed model investigating the relationship between demographic and functional patient variables on cognitive functioning across all four time points. Abbreviations: CERAD = Consortium to Establish a Registry for Alzheimer’s Disease; DAD = Disability Assessment for Dementia (indexing activities of daily living/ADLs); GDS = Geriatric Depression Scale.
What is the added value of the latter findings and how do they contribute to a better understanding of AD?
Upon acknowledging the complex nature of our research questions, it does not come as a surprise that also our findings are complex and far from being comprehensive. However, our prospective multi-center study is among the first to simultaneously examine patients’ demographic, somatic and functional variables as potential predictor variables for disease deterioration (indexed by the CERAD) in patients with early-stage AD. Furthermore, we aimed at elucidating the developmental trajectories of cognitive and non-cognitive functional variables across time (i.e., four assessment points spanning two years). Our findings are significant as they clearly contribute towards a deeper understanding of AD. Above all, our findings might inform future studies targeted at elucidating the complex interplay of patient and caregiver-related predictor variables. For instance, our findings provide a starting point for generating more specific hypotheses. Future studies might utilize objective rather than subjective test measures that -among others- may facilitate cross-national comparisons and, thus, enhance the reliability and generalizability of respective research findings.
Regarding predictive demographic factors, our results identified age and sex (but not education) as significant predictors of cognitive functioning or deterioration. While age is widely recognized as the primary risk factor for developing AD [1,3,4], few studies have reported findings regarding sex differences (see [58] for a comprehensive review). Most interestingly, the findings of our regression analysis disclose that, beyond age, female sex appears to be a risk factor for cognitive deterioration (being considered as a proxy for AD progression). This contribution adds to the current literature by emphasizing the importance of considering sex as a factor in understanding AD risk. On the other hand, education failed to emerge as a predictor of cognitive functioning or deterioration in our prospective study. Though converging evidence suggests that low education increases the risk of dementia (for the findings of a respective meta-analysis, see [59], our results propose that education seems less important for disease progression if other relevant risk factors are considered simultaneously (for similar findings see [57]). These results align with the idea that, although educational attainment contributes to individual differences across the lifespan, it might not necessarily serve as a protective factor against cognitive decline [59,60].
Regarding the somatic risk factors assessed at T1, only atrial fibrillation remained a significant predictor of cognitive functioning at T4 (according to the regression analysis), while none of the other predefined factors (including coronary heart disease, hypercholesterolemia, hypertension, diabetes, systolic and diastolic blood pressure) were found to be significant predictors. Consistent with our findings, there is accumulating evidence supporting the association between atrial fibrillation and cognitive impairment [61,62]. This link persists even in early-onset dementia [63]. Furthermore, several studies have shown that atrial fibrillation not only increases the risk of vascular dementia but also the risk of AD. Such as, the findings of a meta-analysis published in 2021 disclose that patients with atrial fibrillation had a 60% higher risk of incident all-cause dementia, a 40% higher risk of incident AD, and a 70% higher risk of incident vascular dementia [64]. Most likely, the relationship is multifactorial in nature, involving inflammatory, hemodynamic and/or genetic components. As noted by Rivard et al. [62], prospective large-scale studies are required to establish the causal effects of this relationship. Filling this gap, our prospective study found that even in the early stages of the disease, and beyond other somatic risk factors, atrial fibrillation should be considered a significant and clinically relevant risk factor for cognitive deterioration and, consequently, AD progression.
Limitations
Beyond its strengths (i.e., large prospective cohort study comprising a well-defined sample of patients diagnosed with probable and possible AD, targeted at investigating potential predictors of cognitive decline by including somatic and functional patient variables as well as caregiver variables), the present study also has potential limitations. Importantly, our findings might not be readily generalizable to a broader population as our study participants comprised solely Austrian patients with probable and possible AD. Indeed, it has been claimed previously that studies aiming at identifying potential risk factors for dementia incidence and decline are more prevalent in low- and middle-income countries compared to high-income countries (because the former bear a higher risk of potentially modifiable risk factors for dementia [8, also see 3, 60]).
A further potential limitation of the present study are the rather high drop-out rates (i.e., sample size decreasing from initially n = 500 participating patients to n = 169 at the last follow-up visit). Nonetheless, drop-out rates in prospective studies are frequently reported to be considerable [29–31]. A closer look at the reasons for drop-out disclosed that the missing data are not missing at random (MNAR) but are likely correlated with multiple patient and caregiver parameters. Upon acknowledging previous findings revealing that test results can be easily biased by imputing large amounts of MNAR data (i.e., data that are not missing at random [65]), we decided against the multiple imputation of our drop-out data to avoid biasing our results.
Moreover, some of the functional variables rest on patient and caregiver reports (e.g., pain and neuropsychiatric symptoms, respectively) which are subjective in nature and are characterized by low inter-rater reliability. To obtain objective and precise assessment results, technology-assisted assessment tools seem to be a promising avenue for future research endeavors [66,67].
Finally, though our study included quite many potential predictor variables (i.e., somatic and functional patient variables) that were regressed against the cognitive outcome variable (indexed by the CERAD), these variables were not comprehensive. In other words, we were not able to include lifestyle factors such as physical activity, nicotine and alcohol use as potential predictors to our study, although it has been argued that these factors should be considered as modifiable risk factors for dementia [8,9]. Though we did collect patients’ nicotine and alcohol use (but not physical activity) in our prospective multi-center study, these data were very patchy and thus, could not be included in our analyses.
Conclusion
To summarize, our novel findings, derived from a two-year study in early-stage AD, can be categorized into three main points: (i) significant changes in both cognitive and non-cognitive patient variables over this short time period; (ii) highly significant and stable interrelations between caregiver load and functional patient variables (cognitive functioning, ADLs, neuropsychiatric symptoms, and depressive symptoms); and (iii) the best predictors of cognitive functioning at T4 being high age, female sex, atrial fibrillation diagnosis, and poor ADLs at T1. (iv) Additionally, across all assessment points, and beyond age, sex, and ADLs, also time (indicating disease duration/progression) and depressive symptoms significantly contributed to variance explanation of patients’ cognitive functioning (CERAD total score). These findings underscore the importance of a comprehensive treatment approach, considering both patient and caregiver variables, in the diagnosis and treatment of early-stage AD. Such as, our findings disclosing strong correlations between caregiver load and various patient-related measures strongly suggest that the clinical management of early-stage AD should be targeted at the patient-caregiver dyad (instead of solely focusing on the patient). Moreover, our regression findings disclose that a combination of high age, female sex, atrial fibrillation (above and beyond other somatic risk factors), low ADLs and depressive symptomatology should be considered as significant (and partly modifiable) risk factors [8,9] for cognitive deterioration. Hence, we propose that a comprehensive treatment for early-stage AD calls for a multidisciplinary approach, including medical, (neuro)psychological and therapeutic professions.
Acknowledgments
We thank the Austrian Alzheimer’s Society for supporting this study. We wish to thank Ms. Cornelia Kirch for her valuable assistance in preparing the reference list.
References
- 1. Winblad B, Amouyel P, Andrieu S, Ballard C, Brayne C, Brodaty H, et al. Defeating Alzheimer’s disease and other dementias: a priority for European science and society. Lancet Neurol. 2016; 15:455–532. pmid:26987701
- 2. Dubois B, Hampel H, Feldman HH, Scheltens P, Aisen P, Andrieu S, et al. Preclinical Alzheimer’s disease: Definition, natural history, and diagnostic criteria. Alzheimers Dement. 2016;12:292–323. pmid:27012484
- 3.
Gauthier S, Webster C, Servaes S, Morais JA, Rosa-Neto P. World Alzheimer Report 2022: Life after diagnosis: Navigating treatment, care and support. London, England: Alzheimer’s Disease Internat; 2022.
- 4. Dubois B, Villain N, Frisoni GB, Rabinovici GD, Sabbagh M, Cappa S, et al. Clinical diagnosis of Alzheimer’s disease: recommendations of the International Working Group. Lancet Neurol. 2021;20:484–496. pmid:33933186
- 5. Deckers K, Schievink SHJ, Rodriquez MMF, van Oostenbrugge RJ, van Boxtel MPJ, Verhey FRJ, et al. Coronary heart disease and risk for cognitive impairment or dementia: Systematic review and meta-analysis. PLoS One. 2017;12:e0184244. pmid:28886155
- 6. Wolters FJ, Segufa RA, Darweesh SKL, Bos D, Ikram MA, Sabayan B, et al. Coronary heart disease, heart failure, and the risk of dementia: A systematic review and meta-analysis. Alzheimers Dement. 2018;14:1493–1504. pmid:29494808
- 7. Gorelick PB, Scuteri A, Black SE, Decarli C, Greenberg SM, Iadecola C; American Heart Association Stroke Council, Council on Epidemiology and Prevention, Council on Cardiovascular Nursing, Council on Cardiovascular Radiology and Intervention, and Council on Cardiovascular Surgery and Anesthesia. Vascular contributions to cognitive impairment and dementia: a statement for healthcare professionals from the american heart association/american stroke association. Stroke. 2011;42:2672–713. pmid:21778438
- 8. Livingston G, Huntley J, Sommerlad A, Ames D, Ballard C, Banerjee S, et al. Dementia prevention, intervention, and care: 2020 report of the Lancet Commission. Lancet. 2020;396:413–46. pmid:32738937
- 9. Montero-Odasso M, Ismail Z, Livingston G. One third of dementia cases can be prevented within the next 25 years by tackling risk factors. The case "for" and "against". Alzheimers Res Ther. 2020;12:81. pmid:32641088
- 10. Hughes D, Judge C, Murphy R, Loughlin E, Costello M, Whiteley W, et al. Association of blood pressure lowering with incident dementia or cognitive impairment: A systematic review and meta-analysis. JAMA. 2020;323:1934–44. pmid:32427305
- 11. Ma Y, Tully PJ, Hofman A, Tzourio C. Blood pressure variability and dementia: A state-of-the-art review. Am J Hypertens. 2020;33:1059–66. pmid:32710605
- 12. Williamson J. D., Pajewski N. M., Auchus A. P., Bryan R. N., Chelune G., Cheung A. K., et al. Sprint Mind Investigators for the SPRINT Research Group. Effect of intensive vs standard blood pressure control on probable dementia: A randomized clinical trial. JAMA. 2019;321:553–61. pmid:30688979
- 13. Diniz BS, Butters MA, Albert SM, Dew MA, Reynolds CF. Late-life depression and risk of vascular dementia and Alzheimer’s disease: systematic review and meta-analysis of community-based cohort studies. Br J Psychiatry. 2013;202:329–35. pmid:23637108
- 14. Defrancesco M, Marksteiner J, Kemmler G, Fleischhacker WW, Blasko I, Deisenhammer EA. Severity of depression impacts imminent conversion from mild cognitive impairment to Alzheimer’s disease. J Alzheimers Dis. 2017;59:1439–48. pmid:28731429
- 15. Van der Mussele S, Fransen E, Struyfs H, Luyckx J, Mariën P, Saerens J, et al. Depression in mild cognitive impairment is associated with progression to Alzheimer’s disease: a longitudinal study. J Alzheimers Dis. 2014;42:1239–50. pmid:25024328
- 16. Yang L, Deng YT, Leng Y, Ou YN, Li YZ, Chen SD, et al. Depression, depression treatments, and risk of incident dementia: A prospective cohort study of 354,313 participants. Biol Psychiatry. 2023;93:802–9. pmid:36526487
- 17. Husebo BS, Achterberg W, Flo E. Identifying and managing pain in people with Alzheimer’s disease and other types of dementia: A systematic review. CNS Drugs. 2016;30:481–97. pmid:27240869
- 18. Kaufmann L, Moeller K, Marksteiner J. Pain and associated neuropsychiatric symptoms in patients suffering from dementia: Challenges at different levels and proposal of a conceptual framework. J Alzheimers Dis. 2021;83:1003–9. pmid:34366340
- 19. Malara A, De Biase GA, Bettarini F, Ceravolo F, Di Cello S, Garo M, et al. Pain assessment in elderly with behavioral and psychological symptoms of dementia. J Alzheimers Dis. 2016;50:1217–25. pmid:26757042
- 20. Gagliese L, Gauthier LR, Narain N, Freedman T. Pain, aging and dementia: Towards a biopsychosocial model. Prog Neuropsychopharmacol Biol Psychiatry. 2018;87(Pt B):207–15. pmid:28947182
- 21. Defrancesco M, Marksteiner J, Kemmler G, Dal-Bianco P, Ransmayr G, Benke T, et al. Specific neuropsychiatric symptoms are associated with faster progression in Alzheimer’s disease: Results of the prospective dementia registry (PRODEM-Austria). J Alzheimers Dis. 2020;73:125–33. pmid:31744004
- 22. Connors MH, Seeher K, Teixeira-Pinto A, Woodward M, Ames D, Brodaty H. Dementia and caregiver burden: A three-year longitudinal study. Int J Geriatr Psychiatry. 2020;35:250–8. pmid:31821606
- 23. Desai AK, Grossberg GT, Sheth DN. Activities of daily living in patients with dementia: clinical relevance, methods of assessment and effects of treatment. CNS Drugs. 2004;18:853–75. pmid:15521790
- 24. Bramboeck V, Moeller K, Marksteiner J, Kaufmann L. Loneliness and burden perceived by family caregivers of patients with Alzheimer disease. Am J Alzheimers Dis Other Demen. 2020;35:1533317520917788. pmid:32281389
- 25. Brodaty H, Woodward M, Boundy K, Ames D, Balshaw R; PRIME Study Group. Prevalence and predictors of burden in caregivers of people with dementia. Am J Geriatr Psychiatry. 2014;22:756–65. pmid:24012226
- 26. Cheng ST. Dementia caregiver burden: A research update and critical analysis. Curr Psychiatry Rep. 2017;19:64. pmid:28795386
- 27. McKhann G, Drachman D, Folstein M, Katzman R, Price D, Stadlan EM. Clinical diagnosis of Alzheimer’s disease: report of the NINCDS-ADRDA Work Group under the auspices of Department of Health and Human Services Task Force on Alzheimer’s Disease. Neurology. 1984;34:939–44. pmid:6610841
- 28. McKhann GM, Knopman DS, Chertkow H, Hyman BT, Jack CR Jr, Kawas CH, et al. The diagnosis of dementia due to Alzheimer’s disease: recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement. 2011;7:263–9. pmid:21514250
- 29. Li J, Schluchter MD. Conditional mixed models adjusting for non-ignorable drop-out with administrative censoring in longitudinal studies. Stat Med. 2004;23:3489–503. pmid:15505888
- 30. Touloumi G, Babiker AG, Pocock SJ, Darbyshire JH. Impact of missing data due to drop-outs on estimators for rates of change in longitudinal studies: a simulation study. Stat Med. 2001;20:3715–28. pmid:11782028
- 31. Touloumi G, Pocock SJ, Babiker AG, Darbyshire JH. Impact of missing data due to selective dropouts in cohort studies and clinical trials. Epidemiology. 2002;13:347–55. pmid:11964938
- 32. Schmid NS, Ehrensperger MM, Berres M, Beck IR, Monsch AU. The Extension of the German CERAD Neuropsychological Assessment Battery with Tests Assessing Subcortical, Executive and Frontal Functions Improves Accuracy in Dementia Diagnosis. Dement Geriatr Cogn Dis Extra. 2014;4:322–34. pmid:25298776
- 33. Folstein MF, Folstein SE, McHugh PR. "Mini-mental state": A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res. 1975;12:189–98.
- 34. Chandler MJ, Lacritz LH, Hynan LS, Barnard HD, Allen G, Deschner M, et al. A total score for the CERAD neuropsychological battery. Neurology. 2005;65:102–6. pmid:16009893
- 35. Feldman H, Sauter A, Donald A, Gélinas I, Gauthier S, Torfs K, et al. The disability assessment for dementia scale: a 12-month study of functional ability in mild to moderate severity Alzheimer disease. Alzheimer Dis Assoc Disord. 2001;15:89–95. pmid:11391090
- 36. Gélinas I, Gauthier L, McIntyre M, Gauthier S. Development of a functional measure for persons with Alzheimer’s disease: the disability assessment for dementia. Am J Occup Ther. 1999;53:471–81. pmid:10500855
- 37. Yesavage JA, Brink TL, Rose TL, Lum O, Huang V, Adey M, et al. Development and validation of a geriatric depression screening scale: a preliminary report. J Psychiatr Res. 1982–1983;17:37–49. pmid:7183759
- 38. Gauggel S, Birkner B. Validität und Reliabilität einer deutschen Version der Geriatrischen Depressionsskala (GDS). Z Klin Psychol. 1999;28:18–27. German.
- 39. Basler HD, Bloem R, Casser HR, Gerbershagen HU, Griessinger N, Hankemeier U, et al. Ein strukturiertes Schmerzinterview für geriatrische Patienten [A structured pain interview for geriatric patients]. Schmerz. 2001;15:164–71. German. pmid:11810351
- 40. Hadjistavropoulos T, Herr K, Prkachin KM, Craig KD, Gibson SJ, Lukas A, et al. Pain assessment in elderly adults with dementia. Lancet Neurol. 2014;13:1216–27. pmid:25453461
- 41. Rajput K, Ng J, Zwolinski N, Chow RM. Pain management in the elderly: A narrative review. Anesthesiol Clin. 2023;41:671–91. pmid:37516502
- 42. Cummings JL. The Neuropsychiatric Inventory: assessing psychopathology in dementia patients. Neurology. 1997;48(5 Suppl 6):S10–6. pmid:9153155
- 43. Braun M, Scholz U, Hornung R, Martin M. The burden of spousal caregiving: a preliminary psychometric evaluation of the German version of the Zarit burden interview. Aging Ment Health. 2010;14:159–67. pmid:20336547
- 44. Zarit SH, Antony CR, Boutselis M. Interventions with caregivers of dementia patients: comparison of two approaches. Psychol Aging. 1987;2:225–32. pmid:3268213
- 45.
R Core Team. R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing (Version 4.3.0). 2019. https://www.R-project.org/
- 46. Bates D, Mächler M, Bolker B, Walker S. Fitting linear mixed-effects models using lme4. J Stat Softw. 2015;67:1–48.
- 47.
Harrell Jr F. Hmisc: Harrell Miscellaneous. R package (Version 5.0–1). 2023; https://CRAN.R-project.org/package=Hmisc
- 48. Kuhn M. Building Predictive Models in R Using the caret Package. J Stat Softw; 2008;28:1–26.
- 49.
Cohen J. Statistical power analysis for the behavioral sciences. 2nd ed. Hillsdale, NJ: Lawrence Erlbaum Associates, Publishers; 1988.
- 50. Ransmayr G, Hermann P, Sallinger K, Benke T, Seiler S, Dal-Bianco P, et al. Caregiving and caregiver burden in dementia home care: Results from the prospective dementia registry (PRODEM) of the Austrian Alzheimer Society. J Alzheimers Dis. 2018;63:103–14. pmid:29614643
- 51. Edwin TH, Strand BH, Persson K, Engedal K, Selbaek G, Knapskog AB. Neuropsychiatric symptoms and comorbidity: Associations with dementia progression rate in a memory clinic cohort. Int J Geriatr Psychiatry. 2021;36:960–9. pmid:33462872
- 52. Brown PJ, Devanand DP, Liu X, Caccappolo E; Alzheimer’s Disease Neuroimaging Initiative. Functional impairment in elderly patients with mild cognitive impairment and mild Alzheimer disease. Arch Gen Psychiatry. 2011;68:617–26. pmid:21646578
- 53. Pfeffer RI, Kurosaki TT, Harrah CH Jr, Chance JM, Filos S. Measurement of functional activities in older adults in the community. J Gerontol. 1982;37:323–9. pmid:7069156
- 54. Sabbagh MN, Malek-Ahmadi M, Kataria R, Belden CM, Connor DJ, Pearson C, et al. The Alzheimer’s questionnaire: A proof of concept study for a new informant-based dementia assessment. J Alzheimers Dis. 2010;22:1015–21. pmid:20930293
- 55. Skinner J, Carvalho JO, Potter GG, Thames A, Zelinski E, Crane PK, et al. Alzheimer’s Disease Neuroimaging Initiative. The Alzheimer’s Disease Assessment Scale-Cognitive-Plus (ADAS-Cog-Plus): An expansion of the ADAS-Cog to improve responsiveness in MCI. Brain Imaging Behav. 2012;6:489–501. pmid:22614326
- 56.
American Psychiatric Association, DSM Task Force, & American Psychiatric Association. Diagnostic and statistical manual of mental disorders: DSM-5. 5th ed. Washington, DC: American Psychiatric Association; 2013.
- 57. Bowler JV, Munoz DG, Merskey H, Hachinski V. Factors affecting the age of onset and rate of progression of Alzheimer’s disease. J Neurol Neurosurg Psychiatry 1998;65:184–90. pmid:9703168
- 58. Nebel RA, Aggarwal NT, Barnes LL, Gallagher A, Goldstein JM, Kantarci K, et al. Understanding the impact of sex and gender in Alzheimer’s disease: A call to action. Alzheimers Dement. 2018;14:1171–83. pmid:29907423
- 59. Xu W, Tan L, Wang HF, Tan MS, Tan L, Li JQ, Zhao QF, Yu JT. Education and Risk of Dementia: Dose-Response Meta-Analysis of Prospective Cohort Studies. Mol Neurobiol. 2016 Jul;53(5):3113–3123. pmid:25983035
- 60. Lövdén M, Fratiglioni L, Glymour MM, Lindenberger U, Tucker-Drob EM. Education and cognitive functioning across the life span. Psychol Sci Public Interest. 2020;21:6–41. pmid:32772803
- 61. Jacobs V, Cutler MJ, Day JD, Bunch TJ. Atrial fibrillation and dementia. Trends Cardiovasc Med. 2015;25:44–51. pmid:25442732
- 62. Rivard L, Friberg L, Conen D, Healey JS, Berge T, Boriani G, et al. Atrial fibrillation and dementia: A report from the AF-SCREEN international collaboration. Circulation. 2022;145:392–409. pmid:35100023
- 63. Giannone ME, Filippini T, Whelton PK, Chiari A, Vitolo M, Boriani G, et al. Atrial fibrillation and the risk of early-onset dementia: A systematic review and meta-analysis. J Am Heart Assoc. 2022;11:e025653. pmid:35861843
- 64. Papanastasiou CA, Theochari CA, Zareifopoulos N, Arfaras-Melainis A, Giannakoulas G, Karamitsos TD, et al. Atrial fibrillation Is associated with cognitive impairment, all-cause dementia, vascular dementia, and Alzheimer’s disease: A systematic review and meta-analysis. J Gen Intern Med. 2021; 36:3122–35. pmid:34244959
- 65.
Van Buuren S. Flexible imputation of missing data. 2nd edition. Chapman & Hall CRC Press; 2018.
- 66. Maresova P, Tomsone S, Lameski P, Madureira J, Mendes A, Zdravevski E, et al. Technological Solutions for Older People with Alzheimer’s Disease: Review. Curr Alzheimer Res. 2018;15:975–83. pmid:29701154
- 67. Palmdorf S, Stark AL, Nadolny S, Eliaß G, Karlheim C, Kreisel SH, et al. Technology-assisted home care for people with dementia and their relatives: Scoping review. JMIR Aging. 2021;4:e25307. pmid:33470935