Health-Related Quality of Life in Community-Dwelling Older People with Cognitive Impairment: EQ-5D-3L Measurement Properties

Uncorrected Author Proof Journal of Alzheimer’s Disease xx (20xx) x–xx DOI 10.3233/JAD-200806 IOS Press 1 Health-Related Quality of Life in Community-Dwelling Older People with Cognitive Impairment: EQ-5D-3L Measurement Properties 1 2 3 4 Pilar P´ erez-Rosa,b,c,∗, Rafael Vila-Candela,d,e, Salvador Martin-Utrillaa,b,f and Francisco M. Mart´ ınez-Arnauc,g 5 6 aDepartment of Nursing, Universidad Cat´olica de Valencia San Vicente M´ artir, Valencia, Spain7 bGRICPAL Research Group, Universidad Cat´olica de Valencia San Vicente M´ artir, Valencia, Spain8 cFrailty and Cognitive Impairment Research Group (FROG), Universitat de Val`encia, Valencia, Spain9 dDepartment of Obstetrics and Gynecology, Hospital Universitario de la Ribera, FISABIO, Valencia, Spain10 eDepartment of Nursing, Faculty of Nursing and Podiatry, Universitat de Val`encia, Valencia, Spain11 fPalliative Care Unit, Valencia Institute of Oncology, Valencia, Spain12 gDepartment of Physiotherapy, Universitat de Val`encia, Valencia, Spain13 Accepted 17 July 2020 Abstract. 14 Background: Assessing quality of life (QoL) in older people with cognitive impairment is a challenge. There is no consensus on the best tool, but a short, user-friendly scale is advised. 15 16 Objective: This study aimed to assess the psychometric properties of the self-reported and generic EQ-5D (including the EQ index and EQ visual analog scale [VAS]) in community-dwelling older adults with cognitive impairment. 17 18 Methods: Cross-sectional study analyzing the feasibility, acceptability, reliability, and validity of the EQ-5D based on 188 self-administered questionnaires in a sample of community-dwelling older adults with Mini-Mental State Examination (MMSE) scores of 10 to 24 points. 19 20 21 Results: The EQ index was 0.69 (±0.27) and the EQ VAS was 63.8 (±28.54). Adequate measurement properties were found in acceptability and feasibility. Cronbach’s alpha was 0.69. Good validity was observed in the correlation of each dimension of the EQ-5D with geriatric assessment scales. Higher validity was observed for the EQ index compared to the EQ VAS. 22 23 24 Conclusion: The EQ-5D scale could be a good tool for assessing health-related QoL in community-dwelling older adults with cognitive impairment, though it is necessary to assess the dimensions and the EQ index. 25 26 Keywords: Cognitive impairment, community-dwelling, measurement, older people, quality of life27 INTRODUCTION 28 The increasing interest in measuring quality of29 life (QoL) in older people resides in the need to30 ∗Correspondence to: Pilar P´ erez-Ros, Espartero 7, 46010 Valencia, Spain. Tel.: +34 963637412 4262; E-mail: pilar. [email protected]. understand the effectiveness of interventions in pre31 vention and health promotion programs. Resource 32 management is subject to constant innovations, so 33 the implementation of preventive strategies at differ34 ent levels should be evaluated. Interventions targeted 35 at older people should be assessed for their cost36 effectiveness in order to identify those with the 37 strongest capacity to improve QoL and provide value 38 ISSN 1387-2877/20/$35.00 © 2020 – IOS Press and the authors. All rights reserved Uncorrected Author Proof 2P. P ´ erez-Ros et al. / Quality of Life in Older People with Cognitive Impairment for money. Indeed, assessing QoL as an economic39 measure of interventions is a widely used tactic in 40 adults, though less so in older people [1].41 Assessing the effect of health interventions on QoL42 could provide more consistent results than analyzing43 disease outcomes due to the high comorbidity pre44 sented by this population [2]. Quality of life is defined45 as “an individual’s perceptions of their position in46 life, in the context of the culture and value systems in47 which they live, and in relation to their goals, expec48 tations, standards and concerns” [3]. Health-related49 quality of life (HRQoL) can be defined as “how well a50 person functions in their life and his or her perceived 51 wellbeing in physical, mental, and social domains of52 health.” 53 Yet, measuring QoL can be difficult because the54 concept itself is complex. Assessment is character-55 ized by its multidimensionality, with both objective 56 and subjective measures, and multivariate evaluation57 designs are based on the person and their environment58 as well as consideration of the systems perspective,59 which encompasses the multiple environments that60 impact on the person and the increasing participation61 of people with cognitive disability [4].62 Different studies have observed that community63 dwelling older adults have shown higher HRQoL than 64 those in institutions, mainly due to lower comorbid-65 ity and dependency along with greater socialization 66 [5–7]. The most important factors mentioned in the67 literature to improve HRQoL in community-dwelling68 older adults are: cohabitation at home, socialization 69 and social support, independence, and physical activ-70 ity, while the factors that have a negative influence 71 are mainly comorbidities that impair function, pain, 72 depression, falls, frailty, sarcopenia, and dementia73 [8–12].74 Assessing self-perceived HRQoL in older people75 with dementia is generally considered challenging.76 In advanced stages, this syndrome can affect the reli-77 ability of the data collected [13]. Although there are 78 studies showing differences between self-reported79 assessment by the patient with dementia and that by 80 the family caregiver [14], a fact that has led to dis-81 crepant perspectives on the desirability of using a82 proxy, like the primary caregiver or clinicians, ver-83 sus directly obtaining information from the person84 being assessed—which is always preferable when85 conditions allow [15]. Although some authors defend86 the use of proxies, pointing to the good psychome87 tric properties achieved, factors like emotional state 88 or caregiver fatigue can negatively affect a HRQoL 89 assessment [16]. There is a dearth of literature that 90 analyzes HRQoL according to the severity of the 91 dementia or other factors that negatively influence 92 HRQoL, independently of the severity of the condi93 tion, for example a person’s knowledge that they have 94 cognitive impairment [13]. 95 There are different tools for assessing HRQoL in 96 older people, both generic and specific to different 97 pathologies, including dementia. However, there is a 98 need for further research in order to know the validity 99 of dementia-specific instruments in different groups 100 and areas [17, 18]. Moreover, the diversity of the tools 101 makes it difficult to standardize an instrument [19] 102 and compare the results of different studies in the lit103 erature [1]. Using generic scales in populations with a 104 specific pathology would enable comparison of inter105 ventions in different populations and settings, but it is 106 necessary to know their properties in the populations 107 under study. 108 In that regard, the EQ5D has been shown to be a 109 good instrument for assessing HRQoL in community110 dwelling older adults [20] and in institutionalized 111 people with cognitive impairment [21]. This scale 112 is a short and easy-to-use, and it is widely applied 113 in community-dwelling older adults [20, 22]. Diaz114 Redondo et al. [23] analyzed the psychometric 115 properties of the proxy-rated EQ-5D, showing it to be 116 a valid alternative for assessing quality of life in insti117 tutionalized older people with dementia. Ankri et al. 118 [16] also point to the possibility of using the EQ-5D 119 for people with dementia, although different authors 120 indicate the need for more studies that analyze both 121 its properties for assessing HRQoL according to the 122 severity of the dementia and its validity in the absence 123 of a gold standard [16, 24]. 124 The aim of our study was to assess the psychome125 tric properties of the EQ-5D in community-dwelling 126 older adults with cognitive impairment. 127 MATERIALS AND METHODS 128 Study design and participants 129 A cross-sectional study was carried out from 1 130 January 2020 to 13 March 2020. The inclusion 131 criteria were: participants aged 70 years or older; 132 cognitive impairment assessed by family physician 133 after cognitive evaluation with the Mini-Mental State 134 Examination (MMSE), with Cronbach’s alpha 0.90 135 and a score between 10 to 24, indicating cogni136 tive impairment [25, 26]; living independently in the 137 province of Valencia (Spain) and with the ability 138 to read and write. Exclusion criteria were: refusal 139 Uncorrected Author Proof P. P ´ erez-Ros et al. / Quality of Life in Older People with Cognitive Impairment 3 to participate in the study, serious psychiatric prob-140 lems (severe depression subjected to treatment or 141 acute psychosis), or severe cognitive impairment142 (diagnosed previously by a physician), the exis-143 tence of associated disease conditions resulting in144 a life expectancy of under 6 months, blindness, 145 or deafness.146 Sample size description147 A total of 361,575 community-dwelling older148 adults province-wide were included in the population149 census of 2019. The sample size was calculated based150 on this census to estimate a 18.5% incidence of older 151 adults with mild-severe cognitive impairment [27],152 with an alpha error of 5%, precision of 3%, and a153 statistical power of 95%. The final minimum sample 154 of participants required was 165.155 In order to use an adequate sampling frame, we 156 decided to recruit participants over a period of two157 months. The assessment was carried out in primary 158 care centers of Valencia. All people who showed their159 willingness to participate and who met the inclusion160 criteria were included. To encourage participation, 161 posters were hung and flyers distributed; open infor-162 mation sessions were also held for older individuals163 interested in participating in the study. In addition,164 volunteers signed up on a list in each center, and165 their data were recorded using alphanumeric codes 166 identifying the center and the individual.167 Data collection and quality of life assessment168 Personalized interviews were undertaken with169 each participant to perform the geriatric assessment170 and collect data on age and sex as well as cognitive, 171 functional, and emotional variables. Four nurses with 172 at least five years’ experience in primary health care173 centers and nursing homes were in charge of this task, 174 which was part of their routine clinical practice; they 175 were not otherwise involved in the study. All assess176 ments were performed in the morning, after breakfast177 and before lunch, between 9 am and 12 pm.178 Health-related quality of life was assessed using179 the EQ-5D visual analog scale (VAS) and, the EQ180 5D index according to the parameters of the Spanish181 population. In order to provide information on the182 three aspects of HRQoL, the EQ-5D-3L (Levels) 183 was used. First, its descriptive system assesses the184 level of impairment in each of the five dimensions185 included in the scale: mobility, self-care, usual activ-186 ities, pain/discomfort, and anxiety/depression. Each187 dimension has three levels of impairment: no prob188 lems (level 1), some problems (level 2), and extreme 189 problems (level 3). Second, the descriptive response 190 from the EQ-5D can be adapted into an index score. 191 The score ranges from less than 0 to 1 (where 0 is a 192 health state equivalent to death and negative values 193 are worse than death) and 1 is the most positive score 194 (the maximum level of perceived HRQoL according 195 to the five dimensions included on the scale). Finally, 196 the EQ VAS score was obtained by asking the patients 197 to rate their health on a 20 cm vertical scale. The scale 198 ranges from 0–100, where 0 is the ‘worst imaginable 199 health’ and 100 is the ‘best imaginable health’ [22]. 200 The functional and emotional assessment tools 201 were the Barthel Index Basic Activity of Daily Liv202 ing, the Tinetti Balance and Gait Scale, the Yesavage 203 GeriatricDepressionScale(GDS),theLawtonInstru204 mental Activity of Daily Living Scale (IADL), and 205 the VAS pain. In the present study they were used to 206 obtain the information and complete an individual207 ized geriatric assessment for each participant. 208 The Barthel Index (BI) is a scoring technique that 209 measures the patient’s performance in 10 activities of 210 daily life. The items can be divided into a group that 211 is related to self-care (feeding, grooming, bathing, 212 dressing, bowel and bladder care, and toilet use) and 213 a group related to mobility (ambulation, transfers, and 214 stair climbing). The total score ranges from 0 (totally 215 dependent) to 100 (totally independent); scores are 216 awarded in multiples of 5 [28]. The BI takes may 217 vary depending on participants’ tolerance and abili218 ties to self-report for 2 to 5 min and to submit to direct 219 observation for 20 min. 220 The Tinetti scale has a gait score and a balance 221 score. It uses a 3-point ordinal scale of 0, 1 and 2. 222 Gait is scored over 12 and balance is scored over 16 223 for a total of 28 possible points. A total score of less 224 than 19 points indicates a fivefold increased risk of 225 falls, so the lower the total score, the higher the risk 226 of falling [29]. The time to complete is 10 to 15 min. 227 A Short Form GDS consisted of 15 questions. Ten 228 of the 15 items indicate the presence of depression 229 when answered positively, while the other 5 (question 230 numbers 1, 5, 7, 11, 13) are indicative of depres231 sion when answered negatively. Scores of 0 to 4 are 232 considered normal, depending on age, education, and 233 complaints; 5 to 8 indicate mild depression; 9 to 11 234 indicate moderate depression; and 12 to 15 indicate 235 severe depression [30]. This form can be completed in 236 approximately 5 to 7 min, making it ideal for people 237 who are easily fatigued or are limited in their ability 238 to concentrate for longer periods of time. The time 239 Uncorrected Author Proof 4P. P ´ erez-Ros et al. / Quality of Life in Older People with Cognitive Impairment to complete can be up to 20 min, depending on the240 person’s emotional state.241 The Lawton IADL scale measures people’s abil242 ity to perform eight activities (using the telephone,243 shopping for groceries, food preparation, housekeep-244 ing, laundering, self-medicating, transportation, and 245 managing finances). The total score ranges from 0 246 (totally dependent) to 8 (totally independent) [31]. 247 The Lawton IADL scale takes 10 to 15 min to admin248 ister and contains eight items, with a summary score249 from 0 (low function) to 8 (high function). The time250 to complete is 2 to 5 min if self-reported.251 The VAS pain scale [11] is a continuous scale com252 prised of a horizontal or vertical line, usually 10 cm253 in length. For pain intensity, the scale goes from 0 254 (indicating no pain) to 10 (worst imaginable pain) 255 [32]. The time to complete is 2 min. 256 Ethics257 All participants gave their informed consent for258 inclusion before they enrolled in the study. The study 259 was conducted in accordance with the Declaration260 of Helsinki, and the protocol was approved by the261 Clinical Research Ethics Committee of the University262 of Valencia (Valencia, Spain; Project identification 263 code 1060896). 264 Measurement properties265 We analyzed the main measurement properties of266 the QoL instruments, including feasibility, accept267 ability, reliability, and construct validity, according 268 to the criteria set out in Table 1. As there is no gold-269 standard measure for QoL, criterion validity was not 270 appraised [33]. 271 In order to determine the feasibility of the instru272 ment under normal conditions, we analyzed the 273 percentage of missing data [34]. The adequate dis274 tribution of scores among the sample indicates the 275 acceptability [35], including measures of central ten276 dency and floor and ceiling effect [36, 37]. 277 To analyze reliability, we assessed Cronbach’s 278 alpha for internal consistency and the stability of the 279 measure [31]; and the intraclass correlation coeffi280 cient (ICC) to evaluate the stability between the EQ 281 index and the EQ VAS [38]. 282 Validity convergence determines the relationship 283 of the scale with other measures assessing the same 284 construct. We used the correlation coefficient (Pear285 son’s or Spearman’s), considering high correlation, 286 R≥0.50; moderate, R of 0.35 to 0.49; and weak, 287 R≤34, which was assessed according to Feeny et 288 al.’s criteria [39]. Moreover, the correlation between 289 dimensions of EQ and assessment scales was ana290 lyzed. 291 The results of each EQ-5D domain were rated from 292 1 (no problems) to 3 (extreme problems) and com293 pared to the scales assessing the same domains. The 294 Tinetti scale (gait and balance) functional score refers 295 to domain 1 (mobility) in the EQ-5D; the Barthel 296 scale score (ADL) refers to domain 2 (self-care); 297 the Lawton scale (IADL), to domain 3 (usual activ298 ities); the GDS (depression symptoms), to domain 299 4 (anxiety-depression); and the VAS pain scale, to 300 domain 5 (pain). 301 In order to relate the quantitative scales (Tinetti, 302 Barthel, Lawton, GDS, and VAS pain) with the EQ303 5D domains, we classified the validated scores of each 304 scale into three categories, corresponding to the EQ305 5D levels 1 to 3. Thus, for self-care, the distribution on 306 the Barthel Index was: level 1, 65 to 100 points; level 307 Table 1 Measurement properties of QoL instrument Property Criteria Feasibility Percentage of missing data (should be <10%) Percentage of computable data (should be >95%) Mean, median, and standard deviation similar across items (15% maximum divergence) Acceptability Asymmetry and kurtosis should oscillate between –1 and 1 Floor and ceiling effect (percentage of scores in the lower and upper extremes should be <15%) Internal consistency: Cronbach´s alpha (acceptable values ≥0.7) Reliability Stability of the measure: ICC (EQ index and EQ VAS, one-way, random-effects model; acceptable values were ≥0.7) Convergence: a) correlation between EQ index and VAS, and b) correlation between dimensions of EQ and assessment scales Validity Divergence: correlation between EQ VAS and assessment scales (should be low; R ≤0.30) Internal: R2of two linear regressions between EQ index and assessment scales and EQ VAS and assessment scales ICC, intraclass correlation coefficient; EQ index, EuroQol 5 Dimensions 3 Levels Index; EQ VAS, EuroQol Visual Analog Scale. Uncorrected Author Proof P. P ´ erez-Ros et al. / Quality of Life in Older People with Cognitive Impairment 5 2, 35 to 60 points; and level 3, 0 to 30 points for mobil-308 ity. The Tinetti scores were categorized as follows: 309 level 1 (no problems), 20 to 28 points; level 2 (some310 problems), 10 to 19 points; and level 3 (extreme prob-311 lems), 0 to 9 points. For anxiety/depression, the GDS312 scores were transformed as: level 1, 0 to 4 points; 313 level 2, 5 to 10 points; and level 3, 11 to 15 points.314 For usual activities, Lawton scores were ordered as:315 level 1, 6 to 8 points; level 2, 3 to 5 points; and level 3,316 0 to 2 points. For pain/discomfort, VAS scores were 317 distributed into: level 1, 0 to 3 points; level 2, 4 to 7318 points; and level 3, 8 to 10 points.319 Divergent validity refers to the association between 320 the scale and other measures that assess different321 constructs [40]; only the EQ VAS was analyzed 322 since the EQ index is obtained from the computa-323 tion of the dimensions comparable to the assessment324 scales. To assess internal validity, we performed 325 two linear regressions with the EQ index and the326 EQ VAS. In order to understand the relationship327 between the variables included and the value of R2,328 the categorized scales were included into the three329 levels equivalent to the dimensions of the EQ-5D-3L330 (Table 1).331 Statistical analysis 332 The variables are reported as proportions333 and/or means and standard deviation (SD). The334 Kolmogorov-Smirnov test was used to assess nor335 mality, and the Levene test was applied to explore 336 homogeneity of variances for continuous variables 337 (age, MMSE, EQ-5D VAS and EQ-5D index, 338 Barthel Index, Tinetti Index, GDS, Lawton Index,339 and VAS pain). There were no significant outliers.340 The data met the main assumptions of normality,341 so the t-test for independent samples was used to 342 compare means. The chi-squared test was used to343 compare categorical variables (gender).344 RESULTS345 Of the 361 people initially evaluated for eligibil-346 ity, 47.9% (n= 173) were excluded: 6.9% (n= 12)347 declined to participate, 81.5% (n= 141) did not meet348 the selection criteria, and 11.5% (n= 20) were not349 capable of responding to the items on the question350 naire. The final study sample thus comprised 188351 participants (52.1%) with a predominance of women352 (64.9%; n= 122). The participants presented high353 functionality in both IADL and gait and balance, as354 Table 2 Baseline participant characteristics (n= 188) Variables Mean (SD) Min Max Age, y 79.19 (5.18) 70 95 MMSE, points (0–30) 21.18 (3.17) 10 24 Barthel Index, points (0–100) 88.48 (17.27) 0 100 Lawton Index, points (0–8) 6.24 (2.06) 0 8 Tinetti Index, points (0–28) 21.26 (6.84) 0 28 GDS, points (0–15) 3.97 (3.56) 2 15 VAS pain, points (0–10) 3.07 (2.86) 0 10 GDS, Geriatric Depression Scale; MMSE, Mini-Mental State Examination; SD, standard deviation; VAS, visual analog scale. Fig. 1. Distribution of responses on the EuroQol dimensions. Table 3 Feasibility and acceptability properties of the EQ index and EQ VAS (n= 188) EQ index EQ VAS Mean 0.69 63.8 Standard deviation 0.27∗28.54∗ Median 0.74 70 Asymmetry –0.49 –0.76 Kurtosis –0.75 –0.23 Ceiling 29.7†14.6† Floor 5.5 4.8 ∗Divergence >15%; deg effect >10%. well as a low level of depressive symptoms and pain 355 (Table 2). 356 The sample showed good HRQoL, as assessed 357 by the EQ-5D, with over 60% of the participants 358 showing no problems in the dimensions of mobil359 ity, self-care, usual activities, or pain. The anxiety 360 scores showed a larger proportion of participants with 361 some or extreme problems (Fig. 1). Participants pre362 sented good QoL scores on both the EQ index and the 363 EQ VAS, with mean index and VAS scores hovering 364 around 60% to 70% of the maximum possible QoL 365 (Table 3). 366 Uncorrected Author Proof 6P. P ´ erez-Ros et al. / Quality of Life in Older People with Cognitive Impairment Psychometric properties: feasibility and367 acceptability 368 There were few missing data, and just 9.6% 369 (n= 20) of the 208 older adults who were willing to370 participate were unable to respond to the question-371 naire items, so feasibility was adequate. All data were 372 computable.373 After analyzing the items for acceptability, simi-374 lar values were observed in the mean and median,375 while asymmetry and kurtosis were within the rec376 ommended bounds. On the contrary, a ceiling effect377 was observed in both EQ index and EQ VAS, and the378 SD diverged more than 15% (Table 3). 379 Psychometric properties: reliability and validity380 We obtained a value for Cronbach’s alpha of 0.69, 381 very close to the minimum 0.7 acceptable value for382 analyzing internal consistency as a measure of relia383 bility. Moreover, to assess the stability of the measure,384 the ICC for the EQ index and the EQ VAS was cal-385 culated, obtaining a value of 0.01 (95% confidence 386 interval [CI] –0.32 to 0.26 p= 0.461), indicating a387 poor correlation. 388 The correlation between EQ index and EQ VAS389 was R = 0.371, p< 0.001). The correlation between390 the five dimensions of the EQ and the correspond391 ing assessment scales (Tinetti, Barthel, Lawton, GDS,392 and VAS pain), categorized into three levels as indi393 cated in the Methods, was analyzed to determine394 the convergent validity. Significant correlations were395 found in all dimensions: the strongest was for anxiety 396 (>0.7), followed by usual activities (>0.6) and pain397 (>0.5) (Table 4).398 To assess the divergent validity, we analyzed the399 correlations between continuous values of the EQ400 VAS and the geriatric assessment scales. As shown401 in Fig. 2, there were low, statistically significant cor402 relations between the EQ VAS, but the Tinetti and403 Barthel Indexes were more than 0.3. 404 Table 4 Convergent validity. Correlation between EuroQol 5 Dimensions and assessment scales (n= 188) EuroQol 5 Comprehensive geriatric r dimensions assessment scales 1. Mobility Tinetti Index 0.371∗ 2. Self-care Barthel Index 0.340∗ 3. Usual activities Lawton Index 0.628∗ 4. Pain VAS pain 0.504∗ 5. Anxiety Geriatric Depression Scale 0.703∗ VAS, visual analog scale; ∗p< 0.001. Finally, using the EQ index and the EQ VAS as out405 come variables, we performed two linear regressions. 406 We included the scores of the scales categorized in 407 three levels and adjusted for age and sex, as explana408 tory variables. For both the EQ index and the EQ VAS, 409 the model showed a statistically significant relation410 ship, with greater association and R2for EQ index 411 than for EQ VAS (EQ index: F = 26.57; p< 0.001; 412 R = 0.691; R2= 0.477; EQ VAS: F = 11.44; p< 0.001; 413 R = 0.555; R2= 0.311). The Barthel Index (ADL) was 414 the variable that was most strongly (and negatively) 415 correlated with EQ index, whereas the Tinetti Index 416 (gait and balance) was most strongly (and negatively) 417 correlated with EQ VAS (Table 5). 418 DISCUSSION 419 In recent years there has been an increase in 420 the measurement of HRQoL in people with cogni421 tive impairment. There are many specific tools, but 422 despite the current lack of standardization on the best 423 measurement instrument, few studies have analyzed 424 the validity of the generic HRQoL measurement tools 425 [17, 18] in order to be able to compare results with 426 other studies. The present study aimed to analyze the 427 measurement properties of the generic EQ-5D scale 428 in community-dwelling older adults with cognitive 429 impairment. The generic scale has adequate feasibil430 ity, acceptability and reliability. The validity of EQ 431 index was stronger than the EQ VAS. 432 The sample presented a good quality of life, as 433 reflected by the high percentage of participants with434 out problems in the dimensions and in the ceiling 435 effect of the EQ index and EQ VAS although it 436 is lower than other studies on community-dwelling 437 older people with dementia [41]. Older people with 438 cognitive impairment have higher HRQoL when they 439 live in the community compared to an institution. 440 Social factors, dependence, and comorbidity affect 441 this population negatively [42]. 442 Feasibility measurement properties of EQ-5D can 443 be considered acceptable. Despite arguments from 444 some authors indicating the need for the use of a 445 proxy or caregiver intervention in QoL analysis [19], 446 we obtained a good response from the older adults, 447 with few missing data and all data computable. This 448 may be due to the ease of use of the scale as well 449 as the stage of cognitive impairment selected in the 450 inclusion criteria. 451 Acceptability was good, but some aspects did not 452 met criteria such as the ceiling effect and the diver453 Uncorrected Author Proof P. P ´ erez-Ros et al. / Quality of Life in Older People with Cognitive Impairment 7 Fig. 2. Scatter plot between the EQ VAS and the Geriatric Assessment scales. A) Tinetti Index (0–28: lower scores = more dependence in mobility). B) Barthel Index (0–100: lower scores = more dependence in activities of daily living). C) Lawton Index (0–8: lower scores = more dependence in instrumental activities of daily living). D) The Geriatric Depression Scale GDS (0–10: higher scores = more depression). E) The visual analog scale VAS (1–10: higher scores = more pain). *p< 0.001. Uncorrected Author Proof 8P. P ´ erez-Ros et al. / Quality of Life in Older People with Cognitive Impairment Table 5 Multivariate linear regression analysis with EQ-5D Index and VAS as dependent variables (n= 188) EQ index ␤(95% CI) PEQ VAS ␤(95% CI) P Constant 0.83 (0.78, 0.88) <0.001 82.87 (76.91,88.82) <0.001 Barthel –0.16 (–0.26, –0.05) 0.005 –4.4 (–17.13, 8.32) 0.495 Lawton –0.06 (–0.15, –0.04) 0.002 –5.54 (–6.30,7.55) 0.859 Tinetti –0.07 (–0.13, –0.01) 0.024 –13.14 (–20.39, –5.89) <0.001 VAS –0.12 (–0.16, –0.07) <0.001 –10.44 (–15.74, –5.14) <0.001 GDS –0.13 (–0.018, –0.08) <0.001 –11.60 (–17.64, –5.57) <0.001 CI, confidence interval; GDS, Geriatric Depression Scale; VAS, visual analog scale. gence in the SD. This aspect is similar to several454 specific scales for elderly people with dementia,455 where results on these points have also been poor456 [18]. This may be due to the characteristics of this 457 pathology, which impede the correct collection of458 data in some stages.459 Internal consistency was limited in the analyzed460 sample, as evaluated in the reliability analysis by461 means of the Cronbach´s alpha, which showed a mod462 erate value close to 0.7 and a poor ICC between the463 EQ index and EQ VAS. This value is lower than most464 of the specific scales that analyze HRQoL in older465 people with cognitive impairment [18], although bet466 ter values than in other studies using the same scale467 and proxies for the assessment [16, 23]. This fea468 ture could be responsible for a decrease in the values469 obtained in the generic scale. 470 On the other hand, good values were observed for471 construct validity. There was a moderate correlation 472 between the EQ index and the EQ VAS. This may473 be due to the fact that the EQ index is the result of474 the effect of the dimensions that are more stable from 475 one day to the next, while the EQ VAS [22] is more476 sensitive to change, since it responds to how people477 are feeling at the time of the evaluation. While this478 aspect may result in lower validity values for the EQ479 VAS in older people with cognitive impairment [16,480 21], some authors argue that it is a valid measure for 481 evaluating interventions [43].482 Currently there is no tool considered to be the 483 gold standard for assessing HRQoL, and this makes484 it difficult to test the validity of the different instru-485 ments available [23, 24]. Comparing the dimensions 486 of the EQ-5D and the geriatric assessment scales cor-487 responding to each dimension helps us to understand488 the validity based on scales widely used in the geri-489 atric field. In addition, we observed greater validity490 in community-dwelling compared to institutionalized 491 older people [21].492 The divergent validity of the EQ VAS and the493 rating scales was adequate, indicating that the EQ494 VAS is an easily obtainable complementary measure. 495 The correlation found with respect to independence 496 in mobility and self-care indicate that these dimen497 sions are essential in the perception of quality of 498 life in older adults [44]. Finally, a high R2is 499 observed in both regressions (but higher in the EQ 500 index), suggesting that rating scales could be a good 501 tool to analyze the validity of the EQ-5D in this 502 population. 503 Knowing the measurement properties of the EQ504 5D in the community-dwelling older adults with 505 cognitive impairment could encourage clinicians 506 and decision-makers to use a generic scale that is 507 already used in community-dwelling older adults, 508 enabling the comparison of interventions in differ509 ent populations [45]. The measurement properties 510 are acceptable, and although they are lower than the 511 specific tools in some aspects [17, 18], profession512 als should bear in mind that long and difficult-to-use 513 scales could limit the achievement of results in this 514 population [46]. 515 There are many studies on the use of different spe516 cific tools for assessing HRQoL in institutionalized 517 older people [18], but there is less evidence for those 518 dwelling in the community. It is necessary to carry 519 out multicenter studies with different populations in 520 addition to intensifying the study according to stages 521 of severity of dementia. 522 The present study has some limitations. First of 523 all, results may be different if groups are analyzed 524 according to MMSE ranges (moderate or severe) and 525 participants’ level of education has not been ana526 lyzed either. Several measurement properties were 527 not analyzed, such as interobserver validity, as this 528 was a self-administered questionnaire. Likewise, our 529 research team opted not to do a test-retest in order not 530 to make older people return to the health care center 531 another day, even though the EQ VAS assesses the 532 perceived health state in the moment that the scale is 533 used, and this could vary if completed on different 534 days. We also did not analyze different formats of the 535 form, for example, a digital tool. Moreover, we did not 536 consider participants’ comorbidities, frailty, sarcope537 Uncorrected Author Proof P. P ´ erez-Ros et al. / Quality of Life in Older People with Cognitive Impairment 9 nia, or cohabitation at home, which could interfere538 with the results. 539 Conclusions540 We obtained acceptable values on measurement541 properties of the EQ-5D in community-dwelling542 older adults with cognitive impairment. This generic 543 scale could be used for assessing HRQoL in this544 older population in order to compare results in a non-545 pathological population. It is necessary to analyze546 the dimensions and the EQ index since yield better547 measurement properties with respect to the EQ VAS.548 ACKNOWLEDGMENTS549 The authors would like to thank all of the partici-550 pants in this research, without whom the study would551 not have been possible, and the Universidad Cat´ olica 552 de Valencia “San Vicente M´ artir” for its support.553 This research received no external funding. Trans554 lation and publication costs were covered by the555 Universidad Cat´ olica San Vicente M´ artir556 Authors’ disclosures available online (https:// 557 www.j-alz.com/manuscript-disclosures/20-0806r1).558 REFERENCES559 [1] Cleland J, Hutchinson C, Khadka J, Milte R, Ratcliffe J 560 (2019) A review of the development and application of 561 generic preference-based instruments with the older pop-562 ulation. Appl Health Econ Health Policy 17, 781–801.563 [2] Crocker TF, Brown L, Clegg A, Farley K, Franklin M, Simp564 kins S, Young J (2019) Quality of life is substantially worse 565 for community-dwelling older people living with frailty:566 Systematic review and meta-analysis. Qual Life Res 28, 567 2041–2056.568 [3] World Health Organization (WHO) (1993) Study protocol569 for the World Health Organization project to develop a Qual570 ity of Life assessment instrument (WHOQOL). Qual Life571 Res 2, 153–159.572 [4] Verdugo MA, Navas P, G´ omez LE, Schalock RL (2012) The573 concept of quality of life and its role in enhancing human 574 rights in the field of intellectual disability. J Intellect Disabil 575 Res 56, 1036–1045.576 [5] Ramocha LM, Louw QA, Tshabalala MD (2017) Quality577 of life and physical activity among older adults living in 578 institutions compared to the community. S Afr J Physiother 579 73, 342.580 [6] de Medeiros MMD, Carletti TM, Magno MB, Maia LC,581 Cavalcanti YW, Rodrigues-Garcia RCM (2020) Does the582 institutionalization influence elderly’s quality of life? A sys583 tematic review and meta–analysis. BMC Geriatr 20,1-25 584 [7] Heßmann P, Seeberg G, Reese JP, Dams J, Baum E, M¨ uller585 MJ, Dodel R, Balzer-Geldsetzer M (2016) Health-related586 quality of life in patients with Alzheimer’s disease in dif-587 ferent German health care settings. J Alzheimers Dis 51,588 545–561.589 [8] Lu S, Wu Y, Mao Z, Liang X (2020) Association of formal 590 and informal social support with health-related quality of 591 life among Chinese rural elders. Int J Environ Res Public 592 Health 17,1351. 593 [9] P´ erez-Ros P, Mart´ ınez-Arnau FM, Tarazona-Santabalbina 594 FJ (2019) Risk factors and number of falls as determinants of 595 quality of life of community-dwelling older adults. J Geriatr 596 Phys Ther 42, 63–72. 597 [10] S´ anchez-Garc´ ıa S, Gallegos-Carrillo K, Espinel-Bermudez 598 MC, Doubova SV, S´ anchez-Arenas R, Garc´ ıa-Pe˜ na C, Salv` a599 A, Brise˜ no-Fabian SC (2017) Comparison of quality of life 600 among community-dwelling older adults with the frailty 601 phenotype. Qual Life Res 26, 2693–2703. 602 [11] Bonfiglio V, Umegaki H, Kuzuya M (2019) Quality of life 603 in cognitively impaired older adults. Geriatr Gerontol Int 604 19, 999–1005. 605 [12] Henchoz Y, Botrugno F, Cornaz S, B¨ ula C, Charef S, Santos606 Eggimann B, Research Group on the quality of life of older 607 people in cantons of Vaud and Geneva (2017) Determi608 nants of quality of life in community-dwelling older adults: 609 Comparing three cut-offs on the excellent-to-poor spectrum. 610 Qual Life Res 26, 283–289. 611 [13] Easton T, Milte R, Crotty M, Ratcliffe J (2018) An empirical 612 comparison of the measurement properties of the EQ-5D613 5L, DEMQOL-U and DEMQOL-Proxy-U for older people 614 in residential care. Qual Life Res 27, 1283–1294. 615 [14] Jacob L, Han JW, Kim TH, Park JH, Lee SB, Lee JJ, Ryu 616 S-H, Kim S-K, Yoon JC, Jhoo JH, Kim JL, Kwak KP, Moon 617 SW, Kim BJ, Lee DY, Kim KW (2017) How different are 618 quality of life ratings for people with dementia reported by 619 their family caregivers from those reported by the patients 620 themselves? J Alzheimers Dis 55, 259–267. 621 [15] Sands LP, Ferreira P, Stewart AL, Brod M, Yaffe K (2004) 622 What explains differences between dementia patients’ and 623 their caregivers’ ratings of patients’ quality of life? Am J 624 Geriatr Psychiatry 12, 272–280. 625 [16] Diaz-Redondo A, Rodriguez-Blazquez C, Ayala A, 626 Martinez-Martin P, Forjaz MJ, Spanish Research Group on 627 Quality of Life and Aging (2014) EQ-5D rated by proxy in 628 institutionalized older adults with dementia: Psychometric 629 pros and cons. Geriatr Gerontol Int 14, 346–353. 630 [17] Missotten P, Dupuis G, Adam S (2016) Dementia-specific 631 quality of life instruments: A conceptual analysis. Int Psy632 chogeriatr 28, 1245–1262. 633 [18] Perales J, Cosco TD, Stephan BCM, Haro JM, Brayne 634 C (2013) Health-related quality-of-life instruments for 635 Alzheimer’s disease and mixed dementia. Int Psychogeriatr 636 25, 691–706. 637 [19] Chua K-C, B¨ ohnke JR, Prince M, Banerjee S (2019) Health638 related quality-of-life assessment in dementia: Evidence of 639 cross-cultural validity in Latin America. Psychol Assess 31,640 1264–1277. 641 [20] Herdman M, Badia X, Berra S (2001) [EuroQol-5D: A sim642 ple alternative for measuring health-related quality of life in 643 primary care]. Aten Primaria 28, 425–430. 644 [21] P´ erez-Ros P, Mart´ ınez-Arnau FM (2020) EQ-5D-3L for 645 assessing quality of life in older nursing home residents 646 with cognitive impairment. Life 10, 100. 647 [22] EuroQol-EQ-5D-3L. Available online: 648 https://euroqol.org/eq-5d-instruments/eq-5d-3l-about/ 649 (accessed on 17 July 2020). 650 [23] Ankri J, Beaufils B, Novella J-L, Morrone I, Guillemin F, 651 Jolly D, Ploton L, Blanchard F (2003) Use of the EQ-5D 652 among patients suffering from dementia. J Clin Epidemiol 653 56, 1055–1063. 654