Literature Review Resources

The mediating role of depression in the association between disability and quality of life

in Alzheimer’s disease

Mar�ıa G�omez-Gallegoa*, Juan G�omez-Garc�ıab and Ester Ato-Lozanoc

aDepartment of Psychology, Faculty of Health Sciences, Catholic University of Murcia, Murcia, Spain; bDepartment of Quantitative Methods, Faculty of Economics, University of Murcia, Murcia, Spain; cDepartment of developmental psychology, Faculty of

Psychology, University of Murcia, Murcia, Spain

(Received 18 March 2015; accepted 5 July 2015)

Background: An understanding of the determinants of quality of life in Alzheimer’s disease (AD) is required in order to develop effective interventions to promote patients’ well-being. Most studies have pointed out depression, functional ability and environmental factors. However, unmeasured confounders can jeopardize the interpretation of the results. Objectives: To explore the mediating role of depression in the association between functional status and QoL, and establish a procedure to detect confounding variables. Methods: A sample of 192 AD patients and their respective caregivers were recruited from day centers and health care centers in the region of Murcia (Spain). The mediating effect was evaluated using causal mediation analysis. Covariates were introduced into the model in a stepwise fashion and sensitivity analyses were performed to assess the influence of potential confounders. Results: Self-rated depression acted as a partial mediator between functional status and quality of life. The mediating effect was positive and significant even after including both patient- and caregiver-related covariates. Only if confounders explained more than 80% of the residual variance in the mediator or in the outcome, the mediating effects would not be positive. Conclusions: The effect of lack of autonomy on the QoL is mostly explained by the negative consequences on mood status. The sensitivity analysis confirms the robustness of this finding.

Keywords: Alzheimer’s disease; depression; quantitative methods and statistics

Introduction

Alzheimer’s disease (AD) is a neurodegenerative disorder

characterized by deterioration of cognitive and functional

abilities that usually affects both patients and caregivers’

psychological well-being. The world prevalence of AD is

thought to increase in the next decades as life expectancy

does (Rizz, Rosset, & Roriz-Cruz, 2014). Spain has an

aging population and a high prevalence of AD (Pedro-

Cuesta et al., 2009; Serrano, Latorre, & Gatz, 2014). The

care of AD patients has been mainly provided by their

family. Nevertheless, recent changes in the role of

women, a lower number of children per family and an

increase in the number of elderly living alone have led to

alternative solutions, such as ‘rotation’ and cohabitation

arrangement or the use of respite care services (Rivera,

Bermejo, Franco, Morales-Gonz�alez, & Benito-Le�on, 2009). In Spain, contrary to other European countries,

nursing homes are usually rejected by the family care-

givers and considered only as a last resource (Lopez, Los-

ada, Romero-Moreno, M�arquez-Gonz�alez, & Mart�ınez- Mart�ın, 2012). This is not the case of day-care centers that offer both physical and cognitive stimulation to patients

and also respite and support to caregivers (Kwok, Young,

Yip, & Ho, 2013). Current therapies are mainly aimed at

alleviating symptoms and improving quality of life (QoL)

and well-being (Keating & Gaudet, 2012; Moniz-Cook

et al., 2008; Moyle, Fetherstonhaugh, Greben, Beattie, &

AusQoL group, 2015). The assessment of QoL of patients

with AD has presented conceptual and practical chal-

lenges for the past decades (Rabin & Black, 2007). Nowa-

days, there are several instruments for measuring QoL in

AD including patients and proxy versions, and generic

and dementia-specific scales, most of them with good psy-

chometric properties (Bowling et al., 2015; Sch€olzel-Dor- enbos, van der Steen, Engels, & Olde Rikkert, 2007).

Most of the studies on QoL in AD are focused on

ascertaining which factors have higher influence on it.

Overall, the main predictors of QoL are functional

impairment, depressive symptoms, behavioral disturban-

ces and caregivers’ burden (Gomez-Gallego, Gomez-

Amor, & Gomez-Garcia, 2012; Mjørud, Kirkevold,

Røsvik, Selbæk, & Engedal, 2014; Sheehan et al., 2012).

Although in some studies, collinearity among explanatory

variables has been controlled, there can be unmeasured

confounding variables that make difficult the interpreta-

tion of the estimated regression coefficients. In addition,

some of the proposed predictors could be mediating the

relationship between other factors and QoL. Thus, it

seems necessary to explore the relationships between pre-

dictors and QoL and detect the presence of mediation

together with the implementation of methods that increase

confidence in results.

*Corresponding author. Email: [email protected]

� 2015 Taylor & Francis

Aging & Mental Health, 2017

Vol. 21, No. 2, 163�172, http://dx.doi.org/10.1080/13607863.2015.1093603

In this study, the QoL of patients has been considered

as the outcome variable, functional status as the predictor

variable and depressive symptoms as the mediator vari-

able in the relation. There is some agreement in that QoL

implies, besides objective elements, a subjective evalua-

tion of important aspects of one’s life (Brod, Stewart,

Sands, & Walton, 1999; World Health Organization Qual-

ity of Life, 1995). These factors are related with physical

and mental health, functional abilities, participation

in activities, social relationships and financial status

(Lawton, 1994). Depression is a central component of

QoL, not only because is related to psychological well-

being, but also because is associated with more negative

self-perception (health, personal performance and capabil-

ities), and also with more personal and environmental

unmet needs (Houtjes, van Meijel, Deeg, & Beekman,

2011). Besides, depressive AD patients are likely to mis-

interpret neutral faces as sad which could lead to a nega-

tive bias (Weiss et al., 2008). In fact, depression is found

to be the main predictor of QoL at every stage of the dis-

ease in both transversal and longitudinal studies (Gomez-

Gallego et al., 2012; Hoe et al., 2009; Missoten et al.,

2007; Naglie et al., 2011; Tatsumi et al., 2009). Functional

disability mainly results from the progressive cognitive

decline (Brown, Devanand, Liu, & Caccappolo, 2011;

Sut, Ju, Yeon, & Shah, 2004) and predicts caregiver-rated

QoL and, to some extent, QoL self-ratings (Bruvik,

Ulstein, Ranhoff, & Engedal, 2012; Giebel, Sutcliffe, &

Challis, 2015). However, there is some controversy about

the relation between functional disability and depression.

Although, most papers show a weak influence of the

improvement in disability and depressive symptoms

(Bostr€om et al., 2014; Schieman & Plickert, 2007), others suggest a negative effect of depression on functional capa-

bilities, especially in cognitively healthy elderly (Rog

et al., 2014; Zahodne & Tremont, 2013) and in relation

with psychosocial activities (Cipher & Clifford, 2004).

We have two objectives: to test the mediating effect of

depression on the association between functional status

and QoL and to establish a method for the detection of

confounding in linear models. We suggest using sensitiv-

ity analysis coupled with a hierarchical regression analy-

sis to examine the influence of some demographic

covariates in the mediation relation and detect the pres-

ence of unmeasured confounders.

Materials and methods

Sample

Patients and caregivers were recruited from several day-

care centers and health centers in the area of Murcia. The

inclusion criteria for the patients were as follows: (1)

diagnosis of possible or probable AD consistent with the

National Institute of Neurological and Communicative

Disorders and Stroke – Association of Alzheimer Disease

and Related Disorders (NINCDS-ADRDA) criteria

(McKhann et al., 1984); (2) non-severe dementia, defined

as a stage in Global Deterioration Scale (GDS) (Reisberg,

Ferris, De Leon, & Crook, 1982) lower than 6; and (3)

living with a caregiver in the community. Patients with

severe language disturbances were excluded. Caregivers

were defined as the main people providing day-to-day

care. Informed consent was obtained from both patients

and caregivers. The Bioethics Committee of the Univer-

sity of Murcia approved the study.

The final sample consisted of 192 patients and their

caregivers. A majority of the patients were female (n D 118) and their mean age was 75.8 years with standard

deviation (SD) of 6.14. Patients have 5.07 (SD D 2.95) years of education, 104 were married and the rest were

widowed. The stage of the dementia using the GDS was 3

for 61 patients, 4 for 98 patients and 5 for 33 patients. The

mean scores of Geriatric Depression Scale (GDS-15), Bar-

thel Index (BI) and Mini-Mental State Examination

(MMSE) were 5.71 (SD D 2.98), 69.09 (SD D 25.4) and 20.41 (SD D 3.7), respectively. The mean age for the caregivers was 55.9 (SD D 5.24), and 133 of them were women. In 89 of the cases, caregivers were adult children,

in 77 of the cases were spouses and in the rest were no rel-

atives. The attrition rate for this study was zero, and there

were no incomplete data for any patient.

Instruments

Global Deterioration Scale (GDS)

This is a global staging instrument that incorporates

both cognitive and functional aspects of aging and

dementia (Reisberg et al., 1982). It has seven ordinal

stages (1�7) on a scale starting with Stage 1 (no cogni- tive decline) and ending with Stage 7 (very severe cog-

nitive decline). Stages 3�5 correspond to mild cognitive impairment, mild dementia and moderate

dementia, respectively.

Quality of Life in Alzheimer’s Disease Scale (QOL-AD)

This is a 13-item instrument designed to obtain ratings

of patient’s QoL from both patients and caregivers

(Logsdon, Gibbons, McCurry, & Teri, 2002). The scale

reflects the perception of important domains of QoL in

older adults, such as physical health, mood, functional

abilities, family, interpersonal relationships and living

situation. There is also a global item ‘life a as whole.’

Items are scored on a four-point scale, with 1 being

poor and 4 being excellent. Total scores range from 13

(the poorest QoL) to 52 (the highest QoL). The scale

has good reliability and validity when administered to

both patients and caregivers in different cultural set-

tings (G�omez-Gallego, G�omez-Amor, & G�omez-Garc�ıa, 2011; Lin Kiat Yap et al., 2008). Two factors have

been identified: ‘perceive physical health’ and

‘perceived psychological health’ (Gomez-Gallego,

Gomez-Garcia, & Ato-Garcia, 2014). However, most

papers about the psychometric properties of this instru-

ment have been done with mild-to-moderate dementia

patients. Hence, only for this population of patients are

guaranteed the reliability and validity of the scale. In

this study, patients rated their own QoL.

164 M. G�omez-Gallego et al.

Barthel Index (BI)

This is a 10-item scale frequently used to measure perfor-

mance in basic ADL (Mahoney & Barthel, 1965). Items

refer to basic ADL (e.g. grooming, feeding, transfer and

others) and are rated at two to four levels (0, 5, 10, 15

points), being 0 the score corresponding to the highest

dependence in that item (Mahoney & Barthel, 1965). BI is

normally used as a unidimensional instrument, and the

total score is obtained by summing each item score. Total

scores range from 0 (the highest dependence) to 100

(independence). Dependence is classified into five levels

(Shah, Vanclay, & Cooper, 1989): 0�20 (total depen- dence), 21�60 (severe dependence), 61�90 (moderate dependence), 91�99 (mild dependence) and 100 (inde- pendence). The instrument has good internal consistency

and both test�retest and interrater reliability (Formiga, Mascaro, & Pujol, 2005; Sainsbury, Seebass, Bansal, &

Young, 2005).

Geriatric Depression Scale (GDS-15)

GDS is an instrument that could be completed either by

the patient or by the caregiver (Gerety et al., 1994; Yesav-

age, Bronk, Rose, & Lum, 1983). It has been reported to

be valid and reliable when administered to institutional-

ized elders (Conradsson, Rosendahl, et al., 2013). In this

study, a brief version of the scale (15-item version) has

been administered to patients. Several papers note a high

correlation between this version and the original scale

(Alden, Austin, & Sturgeon, 1989; Lesher & Berryhill,

1994). Possible scores range from 0 to 15, with higher

scores indicating more depressive symptoms. A cutoff

point of 5 is used for the diagnosis of a major depressive

episode. In a previous work, we observed a good internal

consistency of GDS-15 in a sample of mild-to-moderate

dementia patients (G�omez-Gallego et al., 2012).

Statistical analysis

There are two main approaches for statistical mediation

analysis, the classical structural equation modeling (SEM)

approach and the causal inference approach. In the classi-

cal approach, the mediation model is a structural model

with two regression equations: one for explaining a medi-

ator M of a predictor X (i.e. X ! M) and another to explain the outcome Y of a predictor X, given the mediator

M (i.e. X ! Y j M). The model may incorporate one or more covariates (Ci) (Baron & Kenny, 1986):

mi D g0 C g1 x1 C g2ci C ei1 (1)

g i D b0 C b1 Xi C b2mi C b3ci C ei2 (2)

The effect of mediation (indirect effect) represents the

changes that X produces on Y transmitted through M. It is

estimated by the product of the coefficients b2 and g1. The direct effect represents the changes that X produces

on Y at a fixed level of the M and is estimated by the

coefficient b1.

The causal inference approach uses the theory of

potential outcomes (Rubin, 2005), which defines a causal

mechanism as a process by which an exposure causes an

outcome in the presence of a mediator. Following Pearl

(2009), identifying a causal mechanism in this approach is

also formulated as a decomposition of the averaged total

effect (ATE) into indirect (average causal mediation

effect, ACME) and direct (average direct effect, ADE)

effects. The identification of ACME and ADE effects

requires sequential ignorability (SI) assumption, which

implies that two ignorability assumptions are made

sequentially. The first assumption implies that, given the

observed pretreatment confounders (e.g. age, gender), the

predictor assignment (e.g. functional disability) is

assumed to be ignorable. Ignorable means statistically

independent of potential outcomes and potential media-

tors. That is to say, there should be no latent confounders

in the X ! Y path (all of the variables that cause both X and Y must be included in the model). In randomized

experimental studies, this assumption is expected to hold.

However, this is not the case of observational research

designs (Ho et al., 2007) like the present study. In this

case, it is necessary to adjust for many pretreatment con-

founders so that the ignorability of predictor assignment

is more credible. The second assumption (ignorability of

the mediator) implies that the observed mediator (e.g.

depression) is independent of all the potential values of

the outcome (QoL) given the actual predictor (e.g. func-

tional status) and pretreatment observed confounders.

This assumption may not hold even in randomized studies

(Imai, Keele, Tingley, & Yamamoto, 2011). In our study,

this assumption implies that among patients who share the

same functional disability and the same pretreatment

covariates (age, gender and others), the depression can be

regarded as if it were randomized. It is not possible to

know for certain if the ignorability of the mediator holds

even after controlling for as many pretreatment confound-

ers as possible (Manski, 2007; Imai, Keele, & Tingley,

2010). The sensitivity analyses are aimed to quantify the

degree to which their empirical findings are robust to a

potential departure of the SI assumption. In other words,

the objective is to assess the sensitivity of the results to an

unobserved confounder that influences both the mediator

and the outcome.

Under the SI assumption, Imai, Keele, and Yamamoto

(2013) proved that the ACME is non-parametrically iden-

tified and provides a valid estimate of the causal media-

tion effect if mediator and outcome are normally

distributed variables. Moreover, they showed that the

causal effects could be estimated as function of the sensi-

tivity parameter r, which represents the correlation

between the residuals of Equations (1) and (2) (Imai et al.,

2013). The SI assumption implies r D 0. Values of r dif- ferent of zero imply deviations of SI assumption pointing

that some confounders are biasing the mediation effect

estimation (Imai et al., 2010). The values of r range from

¡0.9 to 0.9. We can examine how the value of the ACME varies as a function of r and calculate the value of r for

the ACME to be zero or its confidence interval to include

zero (Imai, Keele, Tingley, & Yamamoto, 2010). In these

Aging & Mental Health 165

situations, there is no mediation effect. Another way to

denote the influence of the unobserved confounders is

based on the following decomposition of the error terms

in Equations (1) and (2):

eij D λ1Ui C e;ij

for j D 1, 2, where Ui is an unobserved pretreatment con- founder and the SI is assumed given Ui and Xi. Then, we

can obtain the following coefficients of determination rep-

resenting the percent of residual variance that is explained

by the unmeasured confounders in the mediator (R2�M ) or in the outcome (R2�Y ) (Imbens, 2003):

R2�M D 1¡ Varðe;i1Þ Varðei1Þ D

Varðei1Þ¡Varðe;i1Þ Varðei1Þ

Another interpretation is based on the proportion of the

original variance that is explained by the unobserved con-

founder in the mediator (~R2�M ) or in the outcome (~R 2� Y )

(Imai et al., 2010):

~R2�M D Varðei1Þ¡Varðe;i1Þ

VarðMiÞ D � 1¡R2M

� R2�M

~R2�Y D Varðei2Þ¡Varðe;i2Þ

VarðYiÞ D � 1¡R2Y

� R2�Y

Since there is a relation between r and the coefficients

of determination,

r2 DR2�MR2�Y D~R2M~R2Y=fð1¡R2MÞð1¡R2Y Þg

where R2M and R 2 Y represent the coefficients of determina-

tion from the two regressions (1) and (2). We can obtain

values of ACME for a given value of (R2�M and R 2� Y ) or

(~R2�M and ~R 2� Y ). It is important to know what percent of

variance is explained by the confounder for the ACME to

be zero. The degree in which SI is satisfied is expressed in

terms of the importance of unobserved confounders in

explaining the variance in the mediator and outcome vari-

ables. The variables that were thought to be possible con-

founders should be included in the sensitivity analysis as

covariates.

In this study, we have performed a non-parametric

estimation of the causal effects as described in Imai et al.

(2011). We performed a stepwise hierarchical regression

procedure in order to examine the potential influence of

several covariates on the mediation effects. After the

inclusion of each set of variables, a sensitivity analysis

was performed. First, we introduced a set of covariates

related to patients (gender, marital status, years of educa-

tion and GDS), and after, a set of covariates related to

caregivers (gender and age). We examined the changes in

the direct and mediation effects, and the values of r for

the ACME to be zero. In addition, we calculated the coef-

ficients of determination from the regressions (1) and (2)

and their magnitude for the ACME to be zero.

The causal approach to mediation and sensibility anal-

yses can be applied with Stata (Hicks & Tingley, 2011), R

(Imai et al., 2010) and Mplus (Muthen & Asparouhov,

2015). In this paper, we used the R mediation package

(Tingley, Yamamoto, Hirose, Keele, & Imai, 2014) used

with 1000 Monte Carlo simulations to generate quasi-

Bayesian confidence intervals based on a normal approxi-

mation, as suggested by Imai et al. (2010), and robust esti-

mation of covariance matrix using consistent

heteroskedasticity estimator of R sandwich package

(Lumley & Zeieis, 2013).

Results

The first step of a hierarchical regression analysis revealed

a possible effect of partial mediation among the three vari-

ables with an adjusted R2 of 0.545. The indirect effect was

0.160 (p D 0.000), the direct effect was 0.112 and the pro- portion of mediation was 58.5% (Table 1). The inclusion

of the covariates related to patients increased the adjusted

determination coefficient 0.077 units with respect to the

first model. As seen in Table 1, the mediation effect was

significant and positive (ACME D 0.122; p D 0.010) though the proportion of mediation decreased (39.9%).

Finally, the inclusion of covariates related to caregiver

increased the adjusted R2 to 0.755. We found an indirect

effect of 0.119 (p D 0.010), similar to the estimated value in the model 2 (Table 1). The estimation of direct effect

was ADE D 0.121 (p D 0.000) and the proportion of mediation was nearly 50%.

The results of sensitivity analyses for these models are

summarized in Table 2 and represented in Figures 1�3. The solid line represents the estimated ACME for the

depression mediator for differing values of sensitivity

parameterr. The gray region represents the 95% confi-

dence bands. The dashed line represents the estimated

Table 1. Causal mediation analyses.

Model 1 Model 2 Model 3

Estimate 95% CI p Estimate 95% CI p Estimate 95% CI p

ACME 0.160 0.068�0.261 0.000 0.122 0.026�0.229 0.010 0.119 0.021�0.218 0.010 ADE 0.112 0.004�0.221 0.040 0.182 0.086�0.277 0.000 0.121 0.036�0.207 0.000 TE 0.272 0.128�0.413 0.000 0.305 0.169�0.447 0.000 0.241 0.115�0.371 0.000 PM 0.585 0.307�0.976 0.000 0.399 0.121�0.649 0.010 0.497 0.124�0.795 0.010 ACME: average causal mediation effect; ADE: average direct effect; TE: total effect; PM: proportion of mediation; CI: confidence interval.

166 M. G�omez-Gallego et al.

mediation effect under the SI assumption. This effect is, in

all the models, positive. In the model 1, ACME is zero

whenr is ¡0.7. As shown in Figure 1, the direction of ACME would remain positive unless the value of r was

lower than ¡0.7 (solid line) or if considering the sample variability, this result holds unless r is lower than ¡0.6 (gray region). The products of the determination coeffi-

cients at which ACME is zero are R2�MR 2� Y D 0.490 and

~R2M~R 2 Y D 0.192. That is, ACME would be positive unless

the confounders explained more than 70% of residual var-

iance (square root of 0.490) in the mediator and in the out-

come, though other combinations are possible. Similarly,

ACME would be positive unless the confounders

explained more than 44% of residual variance (square root

of 0.192) in the mediator or in the outcome.

The results of the sensitivity analysis for the

model 2 are shown in Figure 2. The mean value of

r when the ACME is zero is ¡0.75 (confidence interval ¡0.70 to ¡0.80). Similarly, in the model 2, the products of the determination coefficients at which ACME is

zero are R2�MR 2� Y D 0:562 and~R2M~R2Y D 0.176. The ACME

would be positive unless the confounders explained

more than 75% of the residual variance in the mediator

and in the outcome, though other combinations are pos-

sible. Similar reasoning may be applied for the total

variance.

In the final model, the value of r at which ACME is zero is within the interval ranging from ¡0.75 to ¡0.85 (Figure 3). The ACME is guaranteed to be positive unless

r is lower than ¡0.85. In addition, the proportion of the

Table 2. Mediation sensitivity analysis for average causal mediation effect: sensitivity region.

r ACME 95% CI R2�MR 2� Y ~R

2 M~R

2 Y

Model 1

[1] ¡0.75 ¡0.031 ¡0.068�0.005 0.562 0.220 [2] ¡0.70 ¡0.005 ¡0.039�0.027 0.490 0.192 [3] ¡0.65 0.015 ¡0.019�0.049 0.422 0.165 [4] ¡0.60 0.033 ¡0.004�0.360 0.070 1412

Model 2

[1] ¡0.80 ¡0.031 ¡0.064�0.002 0.640 0.201 [2] ¡0.75 ¡0.007 ¡0.031�0.016 0.562 0.176 [3] ¡0.70 0.010 0.014�0.034 0.490 0.033 [4] ¡0.65 0.024 0.005�0.054 0.422 0.132

Model 3

[1] ¡0.80 ¡0.006 ¡0.026�0.120 0.640 0.139 [2] ¡0.75 0.012 ¡0.009�0.033 0.562 0.122 [3] ¡0.70 0.026 ¡0.002�0.055 0.490 0.106

r: sensitivity parameter; ACME: average causal mediation effect; CI: confidence interval; R2�M : proportion of the variance explained by the unobserved confounder in the mediator; R2�Y : proportion of the variance explained by the unobserved confounder in the outcome;~R2M : proportion of unexplained variance that is explained by the unobserved confounder in the mediator;~R2Y : proportion of unexplained variance that is explained by the unobserved confounder in the outcome.

Figure 1. Sensitivity analysis for the model 1.

Figure 2. Sensitivity analysis for the model 2.

Figure 3. Sensitivity analysis for the model 3.

Aging & Mental Health 167

unexplained variance that is explained by unobserved

confounders at which ACME equals zero is 0.640 and the

proportion of the original variance that is explained by

confounders at which ACME is zero is 0.139. Conse-

quently, only if the unmeasured confounders explained

more than 80% of the unexplained variance in the

mediator and in the outcome, the ACME would not be

positive.

Discussion

The present study used a theoretical model to explain the

relation between QoL and two important clinical predic-

tors: depression and functional disability. Moreover, the

validity of the results was assessed using a technique to

detect the presence of unmeasured confounders. Accord-

ing our results, the impact of the loss of autonomy on

patients’ QoL is mostly explained by the consequences on

mood. The mediation effect is positive since higher BI

scores imply lower GDS-15 scores and both changes

imply higher QoL scores. This effect remains significant

and notable regardless patient’s gender, marital status,

educational level, and stage of the disease and caregiver’s

gender and age. Mood and disability interact in a complex

way, depending on several factors as cognitive function,

social support and resilience (Conradsson, Littbrand,

et al., 2013; Hybels, Pieper, & Blazer, 2009; Monaci &

Morris, 2012). In dementia patients, depression mediates

the effect of awareness of dependence on QoL (Woods

et al., 2014). Similarly, depression and social relationships

mediate the relation between perceived health and QoL

(Livingston, Cooper, Woods, Milne, & Katona, 2008).

Furthermore, other personal factors as the individual’s

preferences and needs may mediate the effect of depen-

dence and the social support on QoL (Graff et al., 2007;

Miranda-Castillo et al., 2010). Depression, unmet needs

and social relationships are interrelated (Ball et al., 2010;

Houtjes, van Meijel, Deeg, & Beekman, 2011; Miranda-

Castillo et al., 2010; Miranda-Castillo, Woods, & Orrell,

2010). It is important to improve patients’ social support

and participation in activities in order to increase their

possibilities to live well with the disease (Clare et al.,

2014). Having a good family relationship and being inte-

grated in a supportive social network could strengthen

patients’ resilience to cope with the changes associated to

the disease (Clare, Kinsella, Logsdon, Whitlatch, & Zarit,

2011). There is a reciprocal relationship between the par-

ticipation in social activities and depression in older peo-

ple (Chiao, Weng, & Botticello, 2011; Liu, Leung, & Chi,

2011; Schwarzbac et al., 2013; Siedlecki et al., 2009).

Interestingly, there is an interrelationship between depres-

sion and coping resources (Bjørkløf et al., 2015; de Boer

et al., 2007). There are few studies about strategies aimed

to improve the patients’ capacity to cope with difficulties

in spite of being reduced (Preston, Marshall, & Bucks,

2007; Rickenbach, Condeelis, & Haley, 2015). At least in

mild and moderate stages, AD patients are aware of their

condition and can actively participate in the treatment

(Clare et al., 2014; Johansson, Marcusson, & Wressle,

2015). Recently, promising results have been reported

about the efficacy of the problem adaptation therapy on

depression and disability (Alexopoulos et al., 2011; Kio-

sses et al., 2015). This therapy focuses in enhancing

patients’ coping skills and helps to bypass functional limi-

tations. Other psychological interventions as social sup-

port groups may also be useful to reduce patients’

depression and improve QoL (Leung, Orrell, & Orgeta,

2015). We think that a better understanding of the factors

influencing patients’ QoL should contribute to develop

more efficacious therapies. The model to explain patients’

QoL is highly complex and encompasses multiple predic-

tors and should be tested on longitudinal studies (Clare et

al., 2014). Statistical methods to control confounding vari-

ables as sensitivity analysis are highly recommendable

(Lynch et al., 2008). Notwithstanding, our findings are rel-

evant and add to the literature suggesting that patients’

mood is the main determinant of QoL and mediates the

effect of other predictors. For that reason, an effort should

be done to perform an early detection of depression in AD

and improve the strategies to prevent and treat it.

Methodological connotations

We have tested the consistency of the estimated mediation

effect following the approach proposed by Imai et al.

(2010) of the sensitivity analysis. As shown in the

Figures 1�3 this consistency is high and the existence of confounding is unlikely. The sensitivity coefficient r

reflects the magnitude of bias due to unobserved con-

founders. Although it is interpreted in terms of a range

and has a high degree of subjectivity, it is useful to assess

the degree to which confounding might bias results (Imai

et al., 2011). In this study, we note that the conclusion

about the original direction of the ACME (positive) under

SI is consistent and holds even when considering the sam-

pling variability. Although there is no cutoff value for the

r coefficients that would indicate a problematic level of

susceptibility of the results to confounding, we should

note that the value of r for the ACME to be zero (r D ¡0.7) is very high and the likelihood of the ACME be of the opposite sign is low. That is to say, the conclusion

about the sign of the ACME is plausible given even fairly

large departures from the ignorability of the mediator

(Imai et al., 2010). On the other hand, in order to over-

come the difficulty to interpret the value of r, we also

used the alternative approach recommended by Imai,

Keele, and Yamamoto (2013) to detect the relevance of

confounders. Overall, our results pointed out that an unob-

served confounder should explain more than 80% of the

unexplained residual variance in the GDS-15 (mediator)

and in the QoL (outcome) for the ACME to change the

sign (be negative). According the criteria of Cohen for the

effect size in regression multiple, this is a very high per-

cent (Cohen, Cohen, West, & Aiken, 2003). Since this sit-

uation is unlikely, we must acknowledge that the

estimation of the ACME is very robust to confounding

due to unmeasured confounders. Moreover, the inclusion

of both sets of covariates hardly causes variation in the

percent of variance explained by the unmeasured

168 M. G�omez-Gallego et al.

confounders, which suggests that these do not confound

the mediation relation.

Limitations

This study has several limitations. First, although its

cross-sectional design is used to explain the factors associ-

ated to QoL (Woods et al., 2014; Wilks & Croom, 2008;

Zhang et al., 2014), it does not allow us to establish causal

associations. Second, patients were selected from daycare

centers and health centers. For this reason, they are repre-

sentative only of the community dwelling patients that

use these services. The level of patients’ cognition,

depression and disability, and caregivers’ data are similar

to those reported in other studies with ambulatory patients

and caregivers in Spain (Conde-Sala, Garre-Olmo, Turr�o- Garriga, Vilalta-Franch, & L�opez-Pousa, 2010; Lucas- Carrasco, G�omez-Benito, Rejas, & Brod, 2011).

Conclusions

Unfortunately, dependence is inherent to the progressive

character of the AD. However, this study shows that func-

tional disability does not necessarily lead to reduced QoL.

Its effect highly depends on its consequences on patients’

mood. Strategies to improve patients’ coping skills, social

support and participation in the community may counter-

act the consequences of the dependence. Regarding the

methodological aspects, we note that, to our knowledge,

this is the first clinical research study applying the method

proposed by Imai et al. (2010) to detect confounders.

Regression models applied to non-experimental data are

likely to produce biased estimates of mediation effects.

Consequently, we suggest the use of hierarchical regres-

sion analysis together with sensitivity analysis to assess,

interpret and validate the results of causal mediation anal-

yses performed in these research designs.

Acknowledgements

The authors thank the centers, professionals, patients and care- givers who participated in this study.

Disclosure statement

No potential conflict of interest was reported by the authors.

References

Alden, D., Austin, C., & Sturgeon, R. (1989). A correlation between the Geriatric Depression Scale long and short forms. Journal of Gerontology, 44, 124�125. doi:10.1093/ geronj/44.4.P124

Alexopoulos, G.S., Raue, P.J., Kiosses, D.N., Mackin, R.S., Kanellopoulos, D., McCulloch, C., … Are�an, P.A. (2011). Problem-solving therapy and supportive therapy in older adults with major depression and executive dysfunction: Effect on disability. Archives of General Psychiatry, 68, 33�41. doi:10.1001/archgenpsychiatry.2010.177

Ball, V., Snow, A.L., Steele, A.B., Morgan, R.O., Davila, J.A., Wilson, N., … Kunik, M.E. (2010). Quality of relationships as a predictor of psychosocial functioning in patients with

dementia. Journal of Geriatric, Psychiatry & Neurology, 23, 109�114. doi:10.1177/0891988710363709

Baron, J., & Kenny, D.A. (1986). The moderator-mediator vari- able distinction in social psychological research conceptual, strategic and statistical considerations. Journal of Personal- ity & Social Psychology, 51, 1173�1182. doi:10.1037// 0022-3514.51.6.1173

Bjørkløf, G.H., Engedal, K., Selbæk, G., Maia, D.B., Coutinho, E.S., … Helvik, A.S. (2015). Locus of control and coping strategies in older persons with and without depression. Aging & Mental Health, 8, 1�9. doi:10.1080/13607863 .2015.1040722

Bostr€om, G., Conradsson, M., Rosendahl, E., Nordstr€om, P., Gustafson, Y., & Littbrand, H. (2014). Functional capacity and dependency in transfer and dressing are associated with depressive symptoms in older people. Clinical & Interven- tional Aging, 9, 249�256. doi:10.2147/CIA.S57535

Bowling, A., Rowe, G., Adams, S., Sands, P., Samsi, K., Crane, M., … Manthorpe, J. (2015). Quality of life in dementia, a systematically conducted narrative review of dementia-spe- cific measurement scales. Aging and Mental Health, 19, 13�31. doi:10.1080/13607863.2014.915923

Brod, M., Stewart, A.L., Sands, L., & Walton, P. (1999). Con- ceptualization and measurement of quality of life in demen- tia: The dementia quality of life instrument (DQoL). Gerontologist, 39, 25�35. doi:10.1093/geront/39.1.25

Brown, P.J., Devanand, D.P., Liu, X., & Caccappolo, E. (2011). Alzheimer’s disease neuroimaging initiative. Functional impairment in elderly patients with mild cognitive impairment and mild Alzheimer disease. Archives of Gen- eral Psychiatry, 68, 617�626. doi:10.1001/ archgenpsychiatry.2011.57

Bruvik, F.K., Ulstein, I.D., Ranhoff, A.H., & Engedal, K. (2012). The quality of life of people with dementia and their family carers. Dementia & Geriatric Cognitive Disorders, 34, 7�14. doi:10.1159/000341584

Chiao, C., Weng, L.J., & Botticello, A. L. (2011). Social partici- pation reduces depressive symptoms among older adults: An 18-year longitudinal analysis in Taiwan. BMC Public Health, 11, 292. doi:10.1186/1471-2458-11-292

Cipher, D.J., & Clifford, A. (2004). Dementia, pain, depression, behavioral disturbances, and ADLs: Toward a comprehen- sive conceptualization of quality of life in long-term care. International Journal of Geriatric Psychiatry, 19, 741�748. doi:10.1002/gps.1155

Clare, L., Kinsella, G.J., Logsdon, R., Whitlatch, C., & Zarit, S. H. (2011). Building resilience in mild cognitive impairment and early-stage dementia: Innovative approaches to inter- vention and outcome evaluation. In B. Resnick, L.P. Gwyther, & K.A. Roberto (Eds.), Resilience in aging (pp. 245�259). Springer ScienceCBusiness Media. doi:10.1007/978-1-4419-0232-0

Clare, L., Nelis, S.M., Quinn, C., Martyr, A., Henderson, C., Hindle, J.V., … Victor, C.R. (2014). Improving the experi- ence of dementia and enhancing active life - living well with dementia: Study protocol for the IDEAL study. Health and Quality of Life Outcomes, 12, 164. doi:10.1186/s12955-014- 0164-6

Cohen, J., Cohen, P., West, S.G., & Aiken, L.S. (2003). Applied multiple regression/correlation analysis for the behavioral sciences (3rd ed.). Mahwah, NJ: Erlbaum.

Conde-Sala, J. L, Garre-Olmo, J., Turr�o-Garriga, O., Vilalta- Franch, J., & L�opez-Pousa, S. (2010). Differential features of burden between spouse and adult child caregivers of patients with Alzheimer’s disease: An exploratory compara- tive design. International Journal of Nursing Studies, 47: 1262�1273. doi:10.1016/j.ijnurstu.2010.03.001

Conradsson, M., Littbrand, H., Bostr€om, G., Lindel€of, N., Gus- tafson, Y., & Rosendahl, E. (2013). Is a change in functional capacity or dependency in activities of daily living associ- ated with a change in mental health among older people

Aging & Mental Health 169

living in residential care facilities? Clinical Interventions in Aging, 8, 1561�1568. doi:10.2147/CIA.S53270

Conradsson, M., Rosendahl, E., Littbrand, H., Gustafson, Y., Olofsson, B., & L€ovheim, H. (2013). Usefulness of the Geriatric Depression Scale 15-item version among very old people with and without cognitive impairment. Aging & Mental Health, 17, 638�645. doi:10.1080/13607863.2012 .758231

de Boer, M.E., Hertogh, C.M., Dr€oes, R.M., Riphagen, I.I., Jonker, C., & Eefsting, J.A. (2007). Suffering from dementia � the patient’s perspective: A review of the literature. Inter- national Psychogeriatrics, 19, 1021�1039. doi:http://dx. doi.org/10.1017/S1041610207005765

Formiga, F., Mascaro, J., & Pujol, R. (2005). Inter-rater reliabil- ity of the Barthel index. Age and Ageing, 34, 655�656. doi:10.1093/ageing/afi209

Gerety, M.B., Williams, J.W., Mulrow, C.D., Cornell, J.E., Kadri, A.A., Rosenberg, J., … Long, M. (1994). Perfor- mance of case-finding tools for depression in the nursing home: Influence of clinical and functional characteristics and selection of optimal threshold scores. Journal of Ameri- can Geriatrics Society, 42, 1103�1109. doi:10.1111/j.1532- 5415.1994.tb06217.x

Giebel, C.M., Sutcliffe, C., & Challis, D. (2015). Activities of daily living and quality of life across different stages of dementia, a UK study. Aging & Mental Health, 19, 63�71. doi:10.1080/13607863.2014.915920

G�omez-Gallego, M., Gomez-Garc�ıa, J., & Ato-Garc�ıa, M. (2014) Confirmatory factor analysis of the quality of life in Alzheimer’s disease scale in patients with Alzheimer’s dis- ease. Experimental Aging Research, 40, 266�279. doi:10.1080/0361073X.2014.896664

G�omez-Gallego, M., G�omez-Amor, J., & G�omez-Garc�ıa, J. (2011) Validaci�on de la versi�on espa~nola de la escala QoL- AD en pacientes con enfermedad de Alzheimer, cuidadores y profesionales sanitarios [Validation of the Spanish version of the QoL-AD Scale in Alzheimer disease patients, their carers, and health professionals]. Neurologia, 27, 4�10. doi:10.1016/j.nrl.2011.03.006

G�omez-Gallego, M., G�omez-Amor, J., & G�omez-Garc�ıa, J. (2012). Determinants of quality of life in Alzheimer’s dis- ease perspective of patients, informal caregivers, and profes- sional caregivers. International Psychogeriatrics, 24, 1805�1815. doi:10.1017/S1041610212001081

Graff, M.J., Vernooij-Dassen, M.J., Thijssen, M., Dekker, J., Hoefnagels, W.H., & Olderikkert, M.G. (2007). Effects of community occupational therapy on quality of life, mood, and health status in dementia patients and their caregivers: A randomized controlled trial. Journal of Gerontology: Bio- logical and Medical Sciences, 62, 1002�1009. doi:10.1186/ 1745-6215-15-74

Hicks, R., & Tingley, D. (2011). Causal mediation analysis. The Stata Journal, 11, 609�615.

Ho, D.E., Imai, K., King, G., & Stuart, E.A. (2007). Matching as nonparametric preprocessing for reducing model depen- dence in parametric causal inference. Political Analysis, 15, 199�236. doi:10.1093/pan/mpl013

Hoe, J., Hancock, G., Livingston, G., Woods, B., Challis, D., & Orrell, M. (2009). Changes in the quality of life of people with dementia living in care homes. Alzheimer Disease and Associated Disorders, 23, 285�290. doi:10.1097/ WAD.0b013e318194fc1e

Houtjes, W., van Meijel, B., Deeg, D.J., & Beekman, A.T. (2011). Unmet needs of outpatients with late-life depression: A comparison of patient, staff and carer perceptions. Journal of Affective Disorders, 134, 242�248. doi:10.1016/j. jad.2011.05.052

Hybels, C.F., Pieper, C.F., & Blazer, D.G. (2009). The complex relationship between depressive symptoms and functional limitations in community dwelling older adults: The impact

of subthreshold depression. Psychological Medicine, 39, 1677�1688. doi:10.1017/S0033291709005650

Imai, K., Keele, L., & Tingley, D. (2010). A general approach to causal mediation analysis. Psychological Methods, 15, 309�334. doi:10.1037/a0020761

Imai, K., Keele, L., Tingley, D., & Yamamoto, T. (2010). Causal mediation analysis using R. In H.D. Vinod (Ed.), Advances in social science research using R (pp. 129�154). New York, NY: Springer.

Imai, K., Keele, L., Tingley, D., & Yamamoto, T. (2011). Unpacking the black box of causality learning about causal mechanisms from experimental and observational studies. American Political Science Review, 105, 765�789. doi:10.1017/S0003055411000414

Imai, K., Keele, L., & Yamamoto, T. (2013). Experimental designs for identifying causal mechanisms. Journal of the Royal Statistical Society, 176, 5�51. doi:10.1111/j.1467- 985X.2012.01032

Imbens, G.W. (2003). Sensitivity to exogeneity assumptions in program evaluation. American Economic Review, 93, 126�132. doi:10.1257/000282803321946921

Johansson, M.M., Marcusson, J., & Wressle, E. (2015).Cognitive impairment and its consequences in everyday life: Experien- ces of people with mild cognitive impairment or mild dementia and their relatives. International Psychogeriatrics, 27, 949�958. doi:10.1017/S1041610215000058

Keating, N., & Gaudet, N. (2012). Quality of life of persons with dementia. Journal of Nutrition & Health Aging, 16, 454�456. doi:10.1007/s12603-011-0346-4

Kiosses, D.N., Ravdin, L.D., Gross, J.J., Raue, P., Kotbi, N., & Alexopoulos, G.S. (2015). Problem adaptation therapy for older adults with major depression and cognitive impairment: A randomized clinical trial. JAMA Psychiatry, 72, 22�30. doi:10.1001/jamapsychiatry.2014.1305

Kwok, T., Young, D., Yip, A., & Ho, F. (2013) Effectiveness of day care services for dementia patients and their caregivers. Asian Journal of Gerontology & Geriatrics, 8, 9�15.

Lawton, M.P. (1994). Quality of life in Alzheimer disease. Alz- heimer Disease and Associated Disorders, 8, 138�150.

Lesher, E.L., & Berryhill, J.S. (1994). Validation of the geriatric depression scale � short form among inpatients. Journal of Clinical Psychology, 50, 256�260. doi:10.1002/1097-4679 (199403)50:23.0.CO;2-E

Leung, P., Orrell, M., & Orgeta, V. (2015). Social support group interventions in people with dementia and mild cognitive impairment: A systematic review of the literature. Interna- tional Journal of Geriatric Psychiatry, 30, 1�9. doi:10.1002/gps.4166

Lin Kiat Yap, P., Yen Ni Goh, J., Henderson, L. M., Min Han, P., Shin Ong, K., & Si Ling Kwek, S. (2008). How do Chinese patients with dementia rate their own quality of life? International Psychogeriatrics, 20, 482�493. doi:10.1017/S1041610207006096

Liu, C.P., Leung, D.S., & Chi, I. (2011). Social functioning, pol- ypharmacy and depression in older Chinese primary care patients. Aging & Mental Health, 15, 732�741. doi:10.1080/13607863.2011.562174

Livingston, G., Cooper, C., Woods, J., Milne, A., & Katona, A. (2008). Successful ageing in adversity: The LASER-AD lon- gitudinal study. Journal of Neurology, Neurosurgery and Psychiatry, 79, 641�645. doi:10.1136/jnnp.2007. 126706

Logsdon, R.G., Gibbons, L. E., McCurry, S.M., & Teri, L. (2002). Assessing quality of life in older adults with cogni- tive impairment. Psychosomatic Medicine, 64, 510�519. doi:10.1097/00006842-200205000-00016

L�opez, J., Losada, A., Romero-Moreno, R., M�arquez-Gonz�alez, M., & Mart�ınez-Mart�ın, P. (2012). Factors associated with dementia caregivers’ preference for institutional care. Neu- rologia, 27, 83�89. doi:10.1016/j.nrl.2011.04.003

170 M. G�omez-Gallego et al.

Lucas-Carrasco, R., G�omez-Benito, J., Rejas, J., & Brod, M. (2011). The Spanish version of the dementia quality of life questionnaire: A validation study. Aging & Mental Health, 15, 482�489. doi:10.1080/13607863.2010.543658

Lumley, T., & Zeieis, A. (2013). Sandwich package. R package version 2.3.0. Vienna: R Foundation for Statistical Comput- ing. Retrieved from http//www.R-project.org/

Lynch, K.G., Cary, M., Gallop, R., & Ten Have, T.R. (2008). Causal mediation analyses for randomized trials. Health Services & Outcomes Research Methodology, 8, 57�76. doi:10.1007/s10742-008-0028-9

Mahoney, F., & Barthel, D. (1965). Functional evaluation: The Barthel Index.Maryland State Medical Journal, 14, 61�65.

Manski, C.F. 2007. Identification for prediction and decision. Cambridge, MA: Harvard University Press.

McKhann, G., Drachman, D., Folstein, M., Katzman, R., Price, D., & Stadlan, E.M. (1984). 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. Neu- rology, 34, 939�944. doi:10.1212/wnl.34.7.939

Miranda-Castillo, C., Woods, B., Galboda, K., Oomman, S., Olojugba, C., & Orrell, M. (2010). Unmet needs, quality of life and support networks of people with dementia living at home. Health and Quality of Life Outcomes, 8, 132. doi:10.1186/1477-7525-8-132.

Miranda-Castillo, C., Woods, R.T., & Orrell, M. (2010). People with dementia living alone: What are their needs and what kind of support are they receiving? International Psycho- geriatrics, 22, 607�617. doi:http://dx.doi.org/10.1017/ S104161021000013X

Missotten, P., Ylieff, M., Di Notte, D., Paquay, L., De Lepeleire, J., & Fontaine, O. (2007). Quality of life in dementia: A 2- year follow-up study. International Journal of Geriatric Psychiatry, 22, 1201�1207. doi:10.1002/gps.1814

Mjørud, M., Kirkevold, M., Røsvik, J., Selbæk, G., & Engedal, K. (2014). Variables associated to quality of life among nursing home patients with dementia. Aging & Mental Health, 18, 1013�1021. doi:10.1080/13607863.2014.903468

Monaci, L., & Morris, R.G. (2012). Neuropsychological screen- ing performance and the association with activities of daily living and instrumental activities of daily living in dementia: Baseline and 18- to 24-month follow-up. International Jour- nal of Geriatric Psychiatry, 27, 197�204. doi:10.1002/ gps.2709

Moniz-Cook, E., Vernooij-Dassen, M., Woods, R., Verhey, F., Chattat, R., De Vugt M.,… Interdem group. (2008). A Euro- pean consensus on outcome measures for psychosocial inter- vention research in dementia care. Aging & Mental Health, 12, 14�29. doi:10.1080/13607860801919850

Moyle, W., Fetherstonhaugh, D., Greben, M., Beattie, E., & AusQoL group. (2015). Influencers on quality of life as reported by people living with dementia in long-term care: A descriptive exploratory approach. BMC Geriatrics, 15, 50. doi:10.1186/s12877-015-0050-z

Muthen, B., & Asparouhov, T. (2015). Causal effects in mediation modeling: An introduction with applications to latent varia- bles. Structural Equation Modeling: A Multidisciplinary Jour- nal, 22, 12�23. doi:10.1080/10705511.2014.935843

Naglie, G., Hogan, D.B., Krahn, M., Beattie, B.L., Black, S.E., MacKnight, C., & Tomlinson, G. (2011). Predictors of patient self-ratings of quality of life in Alzheimer’s disease: Cross-sectional results from the Canadian Alzheimer’s Dis- ease Quality of Life (CADQOL) Study. The American Jour- nal of Geriatric Psychiatry, 19, 891�901. doi:10.1097/ JGP.0b013e3182006a7f

Pearl, J. (2009). Causal inference in statistics: An overview. Statistics Surveys, 3, 96�146. doi:10.1214/09-SS057

Pedro-Cuesta, J., Viru�es-Ortega, J., Vega, S., Seijo-Mart�ınez, M., Saz, P., Rodr�ıguez, F.,… del Barrio, J.L. (2009). Prevalence of dementia and major dementia subtypes in Spanish populations:

A reanalysis of dementia prevalence surveys, 1990�2008. BMC Neurology, 9, 55. doi:10.1186/1471-2377-9-55.

Preston, L., Marshall, A., & Bucks, R.S. (2007). Investigating the ways that older people cope with dementia: A qualitative study. Aging & Mental Health, 11, 131�143. doi:10.1080/ 13607860600844572

Rabins, P.V., & Black, B.S. (2007). Measuring quality of life in dementia purposes, goals, challenges and progress. Interna- tional Psychogeriatrics, 719, 401�407. doi:10.1017/ S1041610207004863

Reisberg, B., Ferris, S.H., De Leon, M.J., & Crook, T. (1982). The Global Deterioration Scale for assessment of primary degenerative dementia. American Journal of Psychiatry, 139, 1136�1139. doi:10.1176/ajp.139.9.1136

Rickenbach, E.H., Condeelis, K.L., & Haley, W.E. (2015). Daily stressors and emotional reactivity in individuals with mild cognitive impairment and cognitively healthy controls. Psychology & Aging, 30, 420�431. doi:10.1037/a0038973

Rivera, J., Bermejo, F., Franco, M., Morales-Gonz�alez, J.M., & Benito-Le�on, J. (2009). Understanding care of people with dementia in Spain: Cohabitation arrangements, rotation and rejection to long term care institution. International Journal of Geriatric Psychiatry, 24, 142�148. doi:10.1002/gps.2081

Rizzi, L., Rosset, I., & Roriz-Cruz, M. (2014). Global epidemiol- ogy of dementia: Alzheimer’s and vascular types. Biomedi- cal Research International. Advance online publication. doi:10.1155/2014/908915

Rog, L.A., Park, L.Q., Harvey, D.J., Huang, C.J., Mackin, S., & Farias, S.T. (2014). The independent contributions of cogni- tive impairment and neuropsychiatric symptoms to everyday function in older adults. Clinical Neuropsychology, 28, 215�236. doi:10.1080/13854046.2013.876101

Rubin, D.B. (2005). Causal inference using potential outcomes: Design, modeling, decisions. Journal of the American Statistical Association, 100, 322�331. doi:10.1198/ 016214504000001880

Sainsbury, A., Seebass, G., Bansal, A., & Young, J. (2005). Reli- ability of the Barthel Index when used with older people. Age and Ageing, 34, 228�232. doi:10.1093/ageing/afi063

Schieman, S., & Plickert, G. (2007). Functional limitations and changes in levels of depression among older adults: A multi- ple-hierarchy stratification perspective. Journal of Gerontol- ogy: Social Sciences, 62, S36�S42. doi:10.1093/geronb/ 62.1.S36

Sch€olzel-Dorenbos, C.J., van der Steen, M.J., Engels, L.K., & Olde Rikkert, M.G. (2007). Assessment of quality of life as outcome in dementia and MCI intervention trials: A system- atic review. Alzheimer Disease & Associated Disorders, 21, 172�178. doi:10.1097/WAD.0b013e318047df4c

Schwarzbach, M., Luppa, M., Sikorski, C., Fuchs, A., Maier, W., van den Bussche, H., … Riedel-Heller, S.G. (2013). The relationship between social integration and depression in non-demented primary care patients aged 75 years and older. Journal of Affective Disorders, 145, 172�178. doi:10.1016/ j.jad.2012.07.025

Serrano, J.P., Latorre, J.M., & Gatz, M. (2014). Spain: Promoting the welfare of older adults in the context of population aging. Gerontologist, 54, 733�740. doi:10.1093/geront/gnu010

Shah, S., Vanclay, F., & Cooper B. (1989). Improving the sensi- tivity of the Barthel Index for stroke rehabilitation. Journal of Clinical Epidemiology, 42, 703�709.

Sheehan, B.D., Lall, R., Stinton, C., Mitchell, K., Gage, H., Hol- land, C., … Katz, J. (2012). Patient and proxy measurement of quality of life among general hospital in-patients with dementia. Aging & Mental Health, 16, 603�607. doi:10.1080/13607863.2011.653955

Siedlecki, K.L., Tatarina, O., Sanders, L., Albert, M., Blacker, D., Dubois, B., … Stern, Y. (2009). Comparison of patient and caregiver reports of patient activity participation and its relationship to mental health in patients with Alzheimer’s

Aging & Mental Health 171

disease. The Journals of Gerontology Series B: Psychologi- cal Sciences & Social Sciences, 64B, 687�695. doi:10.1093/geronb/gbp071

Suh, G.H., Ju, Y.S., Yeon, B.K., & Shah, A.A. (2004). Longitu- dinal study of Alzheimer’s disease: Rates of cognitive and functional decline. International Journal of Geriatric Psy- chiatry, 19, 817�824. doi:10.1002/gps.1168

Tatsumi, H., Nakaaki, S., Torii, K., Shinagawa, Y., Watanabe, N., Murata, Y., … Furukawa, T.A. (2009). Neuropsychiatric symptoms predict change in quality of life of Alzheimer dis- ease patients: A two-year follow-up study. Psychiatry & Clinical Neuroscience, 63, 374�384. doi:10.1111/j.1440- 1819.2009.01955.x

Tingley, D., Yamamoto, T., Hirose, K., Keele, L., & Imai, K. (2014). Mediation R package for causal mediation analysis. Journal of Statistical Software, 59, 1�40.

Weiss, E.M., Kohler, C.G., Vonbank, J., Stadelmann, E., Kemm- ler, G., Hinterhuber, H., … Marksteiner, J. (2008). Impairment in emotion recognition abilities in patients with mild cognitive impairment, early and moderate Alzheimer disease compared with healthy comparison subjects. Ameri- can Journal of Geriatrics Psychiatry, 6, 974�980. doi:10.1097/JGP.0b013e318186bd53

Wilks, S.E., & Croom, B. (2008). Perceived stress and resilience in Alzheimer’s disease caregivers: Testing moderation and

mediation models of social support. Aging & Mental Health, 12, 357�365. doi:10.1080/13607860801933323

Woods, R.T., Nelis, S.M., Martyr, A., Roberts, J., Whitaker, C.J., & Markova, I. (2014). What contributes to a good quality of life in early dementia? Awareness and the QoL-AD, a cross- sectional study. Health & Quality of Life Outcomes, 11. doi:10.1186/1477-7525-12-94

World Health Organization Quality of Life. (1995). The World Health Organization Quality of Life assessment (WHO- QOL): Position paper from the World Health Organization. Social Science & Medicine, 41, 1403�1409. doi:10.1016/ 0277-9536(95)00112-K

Yesavage, J.A., Brink, T.L., Rose, T.L., & Lum, O. (1983). Development and validation of a geriatric depression scale a preliminary report. Journal of Psychiatric Research, 17, 37�49. doi:10.1016/0022-3956(82)90033-4

Zahodne, L.B., & Tremont G. (2013). Unique effects of apathy and depression signs on cognition and function in amnestic mild cognitive impairment. International Journal of Geriat- ric Psychiatry, 28, 50�56. doi:10.1002/gps.3789

Zhang, S., Edwards, H., Yates, P., Li, C., & Guo, Q. (2014). Self-efficacy partially mediates between social support and health-related quality of life in family caregivers for demen- tia patients in Shanghai. Dementia & Geriatric Cognitive Disorders, 37, 34�44. doi:10.1159/000351865

172 M. G�omez-Gallego et al.

Copyright of Aging & Mental Health is the property of Routledge and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.