Abstract
Background:
Although early diagnosis has been hypothesized to benefit both patients and caregivers, until now studies evaluating the effect of early dementia diagnosis are lacking.
Objective:
To investigate the influence of early specialist referral for dementia on the risk of institutionalization and functional decline in Activity of Daily Living (ADL).
Methods:
Incident dementia cases were screened in a prospective population-based cohort, the Three-City Study, and initial specialist consultation for cognitive complaint was assessed at dementia diagnosis. Proportional hazard regression and illness-death models were used to test the association between specialist referral and, respectively, institutionalization and functional decline.
Results:
Only one third of the incident individuals with dementia had consulted a specialist for cognitive problems early (36%). After adjustment on potential confounders (including cognitive and functional decline) and competing risk of death, participants who had consulted a specialist early in the disease course presented a higher rate of being institutionalized than those who did not (Hazard Ratio = 2.00, 95% Confidence Interval (CI): 1.09– 3.64). But early specialist referral was not associated with further functional decline (HR = 1.09, 95% CI: 0.71– 1.67).
Conclusions:
Early specialist referral in dementia is associated with increased risk of institutionalization but not with functional decline in ADL. These findings suggest that early care referral in dementia may be a marker of concern for patients and/or caregivers; subsequent medical and social care could be suboptimal or inappropriate to allow patients to stay longer at home.
INTRODUCTION
The prevalence of dementia will approximately double worldwide in the next 20 years [1]. Dementia represents a substantial financial burden on society, one that is similar to the financial burden of heart disease and cancer [2]. The main component of its formal costs comes from institutional care rather than from medical services [2]. Even at the disease onset, institutional care represents the major cost driver [3]. The institutionalization process is complex and causes are likely to be multifaceted involving both contextual and clinical factors. The latter have been widely investigated in dementia and onset of disability in basic Activities of Daily Living (ADL) is a major predictor of institutionalization, leading to 50% increased risk [4, 5]. Recent studies also stated that disability in basic ADL was the most important cost contributor for dementia individuals living at home due to the increased need for home assistance and the monetary valorization of informal care to compensate insufficient professional assistance when autonomy loss in essential activities progresses [3]. A recent study investigating the successive emergence of functional impairments in dementia concluded that a hierarchical structure of basic ADL impairment could be described and disability in bathing and/or dressing could be considered as a first major step in the functional course ofdementia [6].
Since dementia is the leading cause of severe functional disability and institutionalization [7], implementation of timely strategies for dementia patients could have the potential to improve patients’ prognosis, for instance by postponing institutionalization. Nevertheless, dementia is frequently undiagnosed or diagnosed at late stages [8]. Improving early care referral in dementia is therefore the first step in order to provide access to information, to initiate early strategies. Moreover, several surveys highlight that elderly people would prefer to be diagnosed early and to be informed about their diagnosis [9] and screening cognitive performance would be relevant for elderly with mental health conditions [10]. To date, the potential benefits of early care access in dementia have been very poorly investigated [8] and the political drive to screen for dementia in an early manner is matter of controversy [11, 12]. Recently, within an observational study, we demonstrated that early referral to care for dementia did not improve patients’ survival and was even associated with higher mortality compared with individuals who did not consult any physician [13]. Gaining a better understanding of the potential effects of early care referral on different outcomes is critical in the current debate toward cognitive screening and early dementia diagnosis [14]. Therefore, we aimed to approach this issue by investigating different prognosis outcomes. The primary objective was to examine the association between early specialist referral with institutionalization risk. In addition, we sought to determine whether or not early specialist referral influenced further functional decline in basic ADL.
METHODS
Study population
This study was part of the Three-City (3C) Study, a collaborative research program based on a longitudinal cohort of 9,294 subjects aged 65 years or more. Its main objective is to estimate the risk of dementia and cognitive impairment attributable to vascular factors and to define target groups for future preventive strategies. Details of the 3C Study have been reported elsewhere [15]. Trained staff administered standardized questionnaires and performed clinical examinations at baseline and 2, 4, 7, 10, and 12 years after. The study protocol was approved by the ethics committee of the Kremlin-Bicetre University Hospital and Sud-Mediterranée III and all participants provided informed consent.
Diagnosis of dementia
At baseline and at each follow-up visit, trained psychologists assessed cognitive function, and dementia was actively screened. After the neuropsychological examination, participants suspected of having dementia, based on their neuropsychological performances or decline relative to a previous examination, were examined by a neurologist or geriatrician, in the presence of an informant for nearly half of the interviews (47%). For each subject with suspected dementia, the physician who examined the participant documented the evolution and severity of the cognitive disorders and any prior consultation for cognitive problem. After this examination, the final diagnosis of dementia was made by a panel of five highly qualified neurologists, independent of the 3C study investigators, who reviewed all available ancillary information. The diagnosis of dementia was established on the basis of DSM-IV criteria (American Psychiatric Association) [16]. Alzheimer’s disease, possible or probable, was diagnosed according to the NINCDS-ADRDAcriteria [17].
Indicators of care referral or cognitive problem
Care referral was assessed at the follow-up when the dementia diagnosis was posed by the neurologist or geriatrician of the study. It was made by both self-reported information given by the subject and/or the informant and by the inventory of all medications. The data were collected during a face-to-face interview conducted by the neurologist or geriatrician. Individuals were classified into two categories: (i) Referral to specialist for those having consulted a specialist (neurologist, geriatrician or psychiatrist) for their cognitive problems and/or treated with antidementia drugs (cholinesterase inhibitors (ChEIs) or memantine); (ii) No referral to specialist for the others, including those who did not express cognitive problems during any care consultation or those who complained to their general practitioner but did not consult a specialist. In France, specialists are in charge of assessing cognitive performances, discussing the potential diagnosis, assessing risky behavior (wandering, falls, etc.), giving information to patients and families, and monitoring health situations by at least biennial consultations. Nevertheless, implementation of a systematic medical follow-up after the diagnosis is heterogeneous among specialists.
Assessment of covariates
Information on age, gender, educational level (primary level, secondary level, or at least high school), living condition (alone or not), and level of income (<760 Euros, 760– 1500 Euros, or >1500 Euros) was assessed. At the time of dementia diagnosis, depressive symptomatology, diabetes, recent fracture, antipsychotic treatment and clinical characteristics of dementia (etiology, behavioral troubles, cognitive and functional performances, cognitive and functional decline) were considered. Depressive symptomatology was evaluated by the Center for Epidemiologic Studies Depression-Scale (CES-D) [18]; participants were classified as exhibiting a depressive symptomatology if they scored 17 and over for men, or 23 and over for women [19]. Diabetes was defined if antidiabetic treatment was documented or if the participant self-reported diabetes. A recent fracture was considered if it occurred between the follow-up when the diagnosis was made (T0) and the preceding follow up (T-1). Behavioral troubles were considered if at least one recent event (aggressiveness, wandering, or hallucination) had occurred at the time of the dementia screening. As data on behavioral troubles and depressive symptoms were missing for respectively 29 and 23 participants in Bordeaux, an additional missing data category was created in order to keep these participants in multivariate additional analysis on institutionalization. Two etiologies of dementia were distinguished: Alzheimer’s disease possible or probable versus other etiologies. For cognition and functional evaluation, we considered the score at the visit when the dementia diagnosis was made and the decline before this visit. The Mini-Mental State Examination (MMSE) was used as an index of global cognitive performance [20]. Instrumental Activities of Daily Living (IADL) were assessed by the French version of Lawton’s scale [21]. Among the eight IADL, four have been shown to be associated with cognitive performance: telephone use, transportation, medication, and domestic finances [22]. A score (ranging from 4 to 15) was calculated by summing the grades of disability for these activities. For MMSE and IADL scores, decline before dementia diagnosis was the annual rate of evolution between T0 and T-1.
Study samples
Two sets of analyses were performed according to the studied endpoints. Institutionalization was investigated only on the Bordeaux sample because it was not systematically assessed in Montpellier and Dijon. Disability was studied on both Bordeaux and Montpellier samples but not in Dijon where the design was assembled to determine the incidence of dementia; it was not designed to evaluate dementia outcomes. For the two studied outcomes, we focused on incident dementia cases diagnosed at the 2, 4, 7, and 10-year follow-ups.
Of the 184 incident dementia cases living at home at the time of the diagnosis, 5 were excluded because of missing data for initial care referral, leaving 179 for the current analyses on institutionalization.
Regarding the analyses on disability, among the 297 incident dementia cases living at home at the time of the diagnosis, 13 were excluded because of missing data for initial care referral, and 41 because of ADL-disability for dressing and/or bathing at the time of dementia diagnosis, leaving 243 incident dementia cases for the current analyses on disability (Fig. 1).
Assessment of endpoints
Institutionalization was defined as any admission into a nursing home (covering both a private or public establishment that provides maintenance and personal or nursing care for the elderly who are unable to care for themselves properly). In Bordeaux, institutionalizationwas tracked systematically at each follow-up, and, when date of entrance was missing, the institution was contacted. When the precise dates were not available, estimated dates were used based on date of last contact or date of death. Using the Katz scale [23], impairment in basic ADLs were studied by focusing on the maximal level of disability in bathing or dressing as based on a previous work [6]. Disability in bathing and/or dressing was defined if one of these two activities were lost and named “ADL dependency.” Functional decline was studied as the occurrence of an ADL dependency at one time after dementia diagnosis. Mortality was tracked throughout each follow-up. The date of death was systematically confirmed up to the end of 2014 by the native city council or the participant’s general practitioner.
Statistical analyses
Factors potentially associated with the indicator of specialist referral were presented by center applying chi-square tests, analysis of variance, and Kruskal-Wallis test as appropriate.
We used cause-specific Cox proportional hazard regression models to investigate the association between specialist referral and institutionalization, while accounting for death as a competing event. Time to event was thus calculated from the date of follow-up when dementia diagnosis was made until the date of institutionalization, censoring at death for subjects who died before institutionalization. We constructed four subsequent models:
Model 1: Adjustment for age at dementia diagnosis, gender, living condition, educational level and level of incomes;
Model 2: Model 1 + etiology of dementia, MMSE and IADL score;
Model 3: Model 1 + etiology of dementia, MMSE decline and IADL decline before diagnosis and use of antipsychotic drug.
Because of missing data for some health status variables, notably behavioral troubles and depressive symptoms, we constructed an additional model (model 4) by adding to model 3 further adjustments on behavioral troubles, depressive symptoms, diabetes, and presence of a recent fracture. Proportionality of hazards was verified by examining graphs based on Schoenfeld residuals. Interactions between specialist referral and explanatory variables for institutionalization were also investigated.
Cumulative incidence functions of institutionalization among subjects who consulted a specialist for cognitive problems and subjects who did not, adjusted for age, gender, and education, were estimated and plotted from the Fine and Gray subdistribution hazard model which allows direct estimation of cumulative incidence functions in competing risk situations [24].
The relationship between specialist referral and ADL disability (secondary objective) was investigated using a semi-parametric illness-death model accounting for both interval censoring of ADL disability and competing risk with death [25]. Indeed, the time to ADL dependency was interval censored between the visit where ADL dependency was assessed and the visit of dementia diagnosis. Such interval censoring induced some uncertainty of the disability status for subjects who died before being diagnosed with disability. The illness-death model for interval-censored data allows accounting for such uncertainty, and provides much better estimates of hazard ratios than standard cause-specific Cox analysis ignoring interval censoring [26]. Regression coefficients of the illness-death model were estimated using a penalized likelihood implemented in the Smooth Hazard R package [27]. Statistical analyses were performed with SAS statistical software version 9.3 and Smooth Hazard R package. We considered a two tailed p value below 0.05 to be statisticallysignificant.
RESULTS
Description of studied samples
About one third of the incident individuals with dementia had consulted a specialist for cognitive problems. In both samples, participants with initial referral to specialist did not differ from others according to age, gender, level of incomes, etiology of dementia, cognitive abilities, or health status (Table 1). Participants who had consulted a specialist had a higher educational level than the others, but only in Bordeaux. In Montpellier, specialist referral was associated with presence of behavioral troubles and more severe IADL functional decline before diagnosis but not worse functional status at the time of diagnosis. Finally, in both samples, living conditions were also associated with carereferral, those who had not consulted a specialist were more likely to live alone than the others.
Influence of care referral on institutionalization
In the Bordeaux center, of the 179 incident dementia cases, 71 (39.7%) have been institutionalized over the study period and 62 (34.6%) died before being institutionalized. The precise dates of admission into institution were missing for 6 subjects and were then imputed. The mean delay between the follow-up when dementia was screened and institutionalization was 2.5 years. Among the institutionalized participants, respectively 17 (23.9%) and 36 (50.7%) were institutionalized over the first one and two years. About one third of the participants (35.6%) who had not consulted a specialist were institutionalized over the follow-up in comparison with 46.9% among participants who had consulted a specialist. The cumulative incidence of institutionalization accounting for the competing risk of death according to initial specialist referral is shown in Fig. 2. After taking into account sociodemographics, specialist referral was associated with an increased rate of being institutionalized with a Hazard Ratio (HR) of 1.76 (95% CI: 1.05– 2.95; Table 2). Further control on dementia characteristics and medical conditions slightly increased the HR (2.00, 95% CI: 1.09– 3.76, Model 4, Table 2).
Influence of care referral on ADL dependency
In the Bordeaux and Montpellier centers, among the 243 incident dementia cases free of ADL dependency, respectively 38 died before the subsequent follow-up, 58 refused to be assessed over the follow-up and 147 were assessed. Table 3 reports the HR of becoming ADL disabled in dressing and/or bathing in relation to specialist referral. The probability of developing ADL dependency was not higher in participants who had consulted a specialist whatever the model used (HR = 1.09, 95% CI 0.71– 1.67, Model 3; Table 3). Note that the Model 4, which includes medical conditions, did not converge.
DISCUSSION
In this sample of incident dementia participants living at home at the time of the diagnosis and screened through a population-based study, we demonstrated that those who had consulted a specialist early for their cognitive problems presented a nearly twice higher rate of institutionalization in comparison with those who had not consulted a specialist, even when taking into account potential confounders, including cognitive and functional decline. On the contrary, there was no relationship between early care referral and further functional decline in basic ADL. Our results suggest that the increased risk of institutionalization for individuals who had an early referral to a specialist may not be explained by their more severe functionaldecline.
Because the diagnosis process of Alzheimer’s disease and other forms of dementia mainly occurs at relatively late stages, with a delay for appropriate care strategies as a consequence, recent national strategies in Western countries and statements from advocacy groups have strongly promoted early diagnosis and thus early care referral for dementia. Nevertheless, this general consensus is not yet supported by quantitative evidence [8]. Only one study has investigated the potential benefit of early care referral for dementia on patient’s prognosis in “real life.” Indeed, in a previous study based on the 3C cohort, we reported a higher mortality for incident cases who had consulted a specialist, and we assumed that this poorer prognosis could be explained by factors fostering care referral and influencing prognosis [13]. Three potential explanations were discussed: firstly, a potential faster dementia progression owing to care referral; secondly, an incomplete assessment of health conditions with thus potential confusing factors not entirely taken into account; and thirdly the potential role of behavioral troubles. Our present findings may contribute to clarify these assumptions. Indeed, a faster initial progression of dementia for those who had consulted a specialist would have led to a faster functional decline as suggested by the strong association between level of IADL decline at the time of diagnosis and further basic ADL impairment in our results. Secondly, multimorbidity is highly correlated with accelerated functional decline notably in basic ADL for people with dementia even at the disease onset [28]. If participants with early specialist referral had a greater burden in health conditions that were not assessed in our study, they should have reported a more severe functional decline. Thus, the two initial assumptions are not supported by our current findings. The third assumption concerned the potential roles of behavioral troubles which are known to influence recognition of dementia even at its early stage [29], and they affect the frail social equilibrium of a subject living with dementia. Moreover, they add to the complexity of dementia care in clinical practice with detrimental consequences on various prognostic outcomes such as institutionalization, cognitive decline, and death [30, 31]. As our assessment of behavioral troubles was basic and did not take into account the severity or the burden of these troubles on caregivers, we cannot rule out the possibility they have influenced institutionalization without impacting on functional decline.
This study is the first one to investigate the potential effect of early care referral on institutionalization and functional decline among incident dementia cases. Few studies explored the associations between early care strategies or early services use on these outcomes but they were all based on convenience samples of diagnosed patients [32–35]. For instance, O’Connor et al. showed that early intervention did not effect institutionalization rates after dementia diagnosis [35]. Nevertheless, people with moderate to severe dementia living alone had an increased rate of institutionalization in the intervention group. Gaugler et al. demonstrated that dementia individuals using in-home help services early were more likely to delay institutionalization [33]. Nevertheless, these patients were at an advanced stage of the disease. A trial assessing a psychosocial intervention reported no efficacy on institutionalization or functioning in ADL in dementia patients diagnosed at very early stage [34]. Recently, a pragmatic trial added significant knowledge into the understanding of dynamics of care for people with dementia and their family caregivers. Meeuwsen et al. compared the effectiveness of memory clinics with general practitioner care in dementia on various clinical outcomes without any difference in patients’ quality of life, behavior, or daily functioning [36]. In memory clinics, dementia treatment was more often prescribed, and information about dementia was more frequently given to caregivers although it was far from systematic. More surprisingly, no difference was noticed concerning other home services used. Overall, there is no current evidence that early specialist referral or early diagnosis by itself may automatically lead to better clinical care. This fact together with the small magnitude of benefits for drug therapies and the potential harms of labeling may contribute to explain the “dementia diagnosis gap” [37].
In our study, half of the institutionalizations occurred over the two first years and an important difference in cumulative rate of institutionalization by care referral can be observed during this timeframe (see Fig. 2). Even if the observational study design does not allow conclusions about the direct causality effect of early specialist referral, we can assume that attending to a specialist consultation at the disease onset can be consequences of specific and uncovered needs experienced by both patients and caregivers. Difficulties coping with the disease burden is the main reason for institutionalization [38]. Thus, as in France dementia care is mostly provided by memory clinics that were initially implemented mainly as diagnostic settings, the absence of systematic and proactive patient-centered management of dementia after diagnosis could explain our main finding.
The strengths of our study include a long follow-up period of observation and the screening of incident cases of dementia through a population-based cohort allowing us to explore the effect of care referral under real life conditions. The competing risks methods used for both studied outcomes took into account the competitive risk of death, which is of crucial importance when studying dementia outcomes. Finally, thanks to the large assessment of data in the 3C cohort and the longitudinal follow-up design with repeated evaluation of cognitive and functional status, we adjusted our analyses on several clinical and nonclinical parameters documented as predictors of functional decline and institutionalization in dementia. However, some limitations must be discussed. The first one concerned the investigation of functional decline. The delay between two follow-up visits was about 2 years. During this timeframe, the participant could develop ADL dependency and die. This is particularly true if a participant refused, or was not assessed for other reasons at a subsequent follow-up after an initial dementia screening. However, the illness-death model permits us to control this interval-censoring bias. Another limitation is that the observational nature of our study did not allow us to exclude the possibility that unmeasured confounders may have accounted for the findings. Notably, loss of social and emotional support or occurrence of behavioral troubles may have affected institutionalization and is not easily assessable for such frailpopulation.
In conclusion, our findings demonstrated an increased rate of institutionalization for participants with dementia who had been referred early to a specialist but without any association with further functional decline. The observed effect in this group could be explained by suboptimal medical and social follow-up to organize dementia care following diagnosis. Our findings highlight the need of improving dementia care by developing an integrated approach covering multidimensional needs assessment, close collaboration between primary care, social and specialized care.
Footnotes
ACKNOWLEDGMENTS
The Three-City study is conducted under a partnership agreement between the Institut National de la Santé et de la Recherche Médicale (INSERM), the University Bordeaux 2 Victor Segalen and Sanofi-Aventis. The Fondation pour la Recherche Médicale funded the preparation and initiation of the study. The Three-City study is also supported by the Caisse Nationale Maladie des Travailleurs Salariés, Direction Générale de la Santé, MGEN, Institut de la Longévité, Conseils Régionaux d’Aquitaine et Bourgogne, Fondation de France, Ministry of Research-INSERM Programme “Cohortes et collections de données biologiques,” Agence Nationale de la Recherche ANR PNRA 2006 and LongVie 2007, and the “Fondation Plan Alzheimer” (FCS 2009-2012).
