Psychometric validation of the Hospital Anxiety and Depression Scale (HADS) in community-dwelling older adults

Depression is prevalent among older adults (≥ 70 years) worldwide, with a point estimate of depressive disorders of 5.4%, including both MDD and CSDS [15]. Furthermore, a recent review based on 20 studies, among which 14 were from community-dwelling settings, found a pooled point prevalence of 13.3% for MDD among older adults [16]. Two recent systematic reviews primarily based on studies of community-dwelling older adults found a pooled point prevalence of 28.4% and 31.7% for CSDS, with a large variation between countries [17, 18]. These two reviews explained the large variety of CSDS estimates by cultural differences along with differences in sample characteristics, methodology, and screening tools used in the included original studies [17, 18].

Compared to younger age groups, older adults report a higher frequency of anxiety [12, 19]. In 2019, the estimated prevalence of anxiety disorders (a combined estimate of all subtypes) among adults aged ≥ 70 years was 4.4% [15]. A recent review among community-dwelling adults ≥ 55 years old revealed a pooled prevalence of anxiety disorders and anxiety symptoms of 5.4% and 7.9%, respectively [20]. Generalized anxiety disorder (GAD) is the most prevalent, leading to a high risk of death, even higher when accompanied by depression [21]. Similar to depression, anxiety symptoms are less characteristic among older adults than in younger age groups [20]; for instance are worries about health, sleep disturbances, and reduced reassurance-seeking behaviors more common symptoms of anxiety in older populations [19].

Moreover, compared to younger adults, older people are less likely to report and seek help for their symptoms, have less knowledge regarding anxiety disorders and available treatment, and face barriers to treatment such as stigma, cost, transportation, and mobility [22]. Ageism, making older people regularly experience prejudice and discrimination in health care, is another reason for not seeking treatment. Hence, ageism might contribute to heightened levels of anxiety and depression [23]. Furthermore, multimorbidity may lead to overlooking clinically relevant symptoms of depression and anxiety. To uncover symptoms of depression and anxiety among older community-dwelling people, easy access to and use of reliable and valid screening measure are needed.

Possibly, the lack of treatment for clinically relevant symptoms of depression and anxiety represents an alarming concern to the healthcare system and society; health politics highlights the need for health promotion interventions to keep older people in their own homes as long as possible. Undetected issues of depression and anxiety may cause reduced mental and physical functionality and increased risk of dementia and other diseases [1, 3‐5], all of which trigger a need for professional health care.

In many older people, depressive symptoms can be difficult to distinguish from age-related symptoms such as sleep disturbance, psychomotor changes, concentration disturbance, changes in nutrition and digestive functioning, and fatigue [4]. Furthermore, anxiety and depression may co-exist [23]. Hence, differentiating between anxiety and depressive symptoms is complicated, caused by overlapping and coinciding symptoms. For example, fatigue, physical symptoms, and negative ruminations have the same clinical features in both diseases [23].

The Hospital Anxiety and Depression Scale (HADS) is commonly used in epidemiological research to estimate the prevalence of clinically significant anxiety symptoms and CSDS among adults [17] and in clinical settings to detect clinically relevant symptoms [1, 24, 25]. HADS was developed to detect clinically relevant symptoms of both anxiety and depression without including any physical symptoms. Thus, the HADS was developed to distinguish between depression and anxiety symptoms. However, since anxiety and depression symptoms may co-exist, we expect these two constructs to correlate to some extent.

The HADS is widely used and tested with satisfactory psychometric properties [26, 27]. However, only a handful of studies have evaluated HADS using confirmative factor analysis (CFA) in community-dwelling, non-clinical samples of older adults [26, 28, 29]; these studies support the original two-factor structure. Problems with items loading on both anxiety and depressive symptoms (cross-loadings) are reported, though [26]. Furthermore, a psychometric evaluation of HADS in a clinical sample of veterans utilizing CFA reported a three-factor structure showing the best fit [30]; the third factor was entitled “dealing with the inability to feel pleasure.” Thus, the dimensionality of HADS is questioned [31].

Initially, the HADS was developed for adults receiving treatment for physical health problems in general hospitals [31]. Still, HADS is used among community-dwelling and older people; evidence suggests a lower cut-off for older adults than among younger populations [32]. This may indicate that HADS, developed for younger hospitalized age groups, might have psychometrical traits, making it less suitable for older adults [20].

Utilizing principal component analysis (PCA), three studies have assessed the dimensionality of the Norwegian version of HADS: one assessed older adults admitted to somatic hospitals [33]. In comparison, the two others did not explicitly study older adults [34, 35]. Nevertheless, using only PCA for psychometrical evaluation implicates several limitations. According to the literature, the CFA approach is needed to achieve a robust test of a scale's dimensionality, composite reliability, and construct validity [36]. One study tested the psychometrics of HADS among older adults in nursing homes [37], reporting that several items revealed low reliability and validity in this population. The authors related these findings to characteristics of the nursing home population, such as several chronic diseases, symptom severity, losses of functionality, loneliness, and dependency of care [38]. The characteristics of community-dwelling older people ≥ 70 years living at home differ compared to the nursing home population. Therefore, a psychometric study of HADS among older adults aged 70 + living at home is required. Concerning early detection and treatment of anxiety and depressive symptoms among community-dwelling older people, a valid and reliable scale is highly needed.

To date, the dimensionality, internal consistency, construct validity, and homogeneity of the Norwegian version of the HADS have not been assessed using CFA among community-dwelling older adults ≥ 70 years. Therefore, this study aims to examine the psychometric properties of the HADS among older adults ≥ 70 living at home in Norway; dimensionality, reliability, and construct validity are interrelated measurement properties and are thus investigated in this study. The hypotheses are:

The hypotheses are based on theory and evidence [1, 17, 23‐25]; we hypothesized that the original measurement model of HADS shows a good fit to our data [26, 27], good reliability and validity [26, 27, 31], comprises two distinct concepts [23] which correlate positively with each other and negatively with QoL. It is rational that when anxiety and depression increase, QOL decreases. To sum up, H_1, H₄, and H₅ are chosen according to the theoretical assumption that HADS with anxiety and depression are two distinct concepts, distinctively different from each other and QOL, and therefore, we assume that these concepts provide good reliability and validity (H₂ and H₃)_.

During 2017–2019, persons aged ≥ 70 were recruited from one randomly selected district in Trondheim, county of Trøndelag in Central Norway, as a part of the fourth wave of The Trøndelag Health Study (HUNT). HUNT is a population-based cross-sectional study [39]. The HUNT study comprises questionnaires, e.g., HADS, clinical measurements, and collections of biological samples. Older age is defined in HUNT as 70 years and above and is linked to retirement for public sector employees, which is 70 years.

In total, 4667 community-dwelling inhabitants living at home in Trondheim were invited to participate, of whom a total of 1486 (response rate of 31.8%) persons 70 years or older (55.3% women and 44.7% men) participated in the regular protocol and responded to the HADS questionnaire and demographical questions (Q1 and Q2). These participants also underwent a comprehensive clinical evaluation.

The inclusion criteria for the present study were: (1) living at home, (2) aged 70 years or older, (3) without dementia, and (4) having responded to all 14 HADS items since this is considered best practice utilizing CFA. In total, 126 participants were diagnosed with a dementia diagnosis, and 170 participants lacked one or more responses on HADS; all of these were excluded from the analyses, giving an effective sample N = 1190.

Sociodemographic characteristics measured were age, sex (male, female), cohabiting status (“no, I live alone” or “yes, with a spouse/partner”), and educational attainment (primary and lower secondary school 9–10 years, academic or vocational school 1 or 2 years, academic or vocational school 3 years, vocational school/apprentice 3–4 years, college or university < 4 years, college, or university ≥ 4 years).

Medical conditions were reported using self-reported items regarding a history of asthma, diabetes, and heart attack. Functional impairment in daily life > 1 year was self-reported using one item (no or yes). Lastly, care support last year in terms of home care, in-home nursing care, and/or hospitalization in nursing homes were self-reported (yes/no).

Global quality of life (QoL) was assessed with one item: “Thinking about your life at the moment, would you say that you by large are satisfied with life, or are you mostly dissatisfied?” The item was scored on a 7-point scale ranging from very satisfied to very dissatisfied.

The HADS consists of 14 items, including subscales for anxiety (HADS-A; seven items) and depression (HADS-D; seven items). The items are scored on a four-point scale ranging from totally disagree to agree totally. Each item is rated from 0–3, where higher scores indicate more severe anxiety and/or depressive symptoms. The maximum score on each subscale is 21, ranging from 0–7 (normal), 8–10 (mild disorder), 11–14 (moderate disorder), and 15–21 (severe disorder) [31].

To increase acceptability and prevent individuals from feeling tested for mental disorders, symptoms of severe psychopathology have been excluded, which makes HADS more sensitive to milder psychopathology [37, 40]. According to the International Classification of Diseases (ICD-10), five of seven items in HADS-D focus on lack of positive feelings and cover only two of three main criteria for depression; physical symptoms such as loss of energy, sleep- and appetite disturbances are not covered.

The analyses were conducted using the IBM Statistical Package for the Social Sciences Version 28 software [41] and the Stata 17 software package [42]. Confirmatory Factor Analysis (CFA) represents a more accurate evaluation of the psychometric properties of the scales used. In this study, the model fit adequacy was assessed by χ²-statistics and conventional fit indices: χ²-statistics, the Root Mean Square Error of Approximation (RMSEA), and the Standardized Root Mean Square Residual (SRMR) with values < 0.10 are acceptable, and values ≤ 0.05 indicates a good fit [43, 44]. Further, the Comparative Fit Index (CFI) and the Tucker-Lewis Index (TLI) with acceptable fit at 0.95 and a good fit at 0.97 [43‐46]. According to Hair et al. [47], skewness and kurtosis should be below an absolute value of 2.0 (standardized); this was the case for all items (data not shown), indicating that both skewness and kurtosis were significant. Therefore, the Satorra-Bentler corrected χ² which is the correct asymptotic mean even under non-normality, is reported [48].

CFA is well known to be sensitive to sample size [49]; the larger the samples, the bigger the chi-square. A consequence may be that well-fitting models are rejected because the chi-square is too high, resulting from a larger sample size more than the real model fit. The present N is 1190, which is considered large. Therefore, we planned to randomly split the dataset into two equally sized parts to check for model fit in each part, termed Sample 1 and Sample 2; all models involved in this study were tested in both samples. Consequently, the original two-factor solution was tested in the total sample as well as in the two sub-samples.

Since we aimed to find the best fitting model, i.e., a model that represents the observed data of older home-dwelling adults in the best manner, we first tested the original model and then, after, other models based on the findings in this original model. Thus, as stated in the aim section, this study investigates the dimensionality, reliability, and construct validity of the HADS among community-dwelling adults 70 years and older; dimensionality, reliability, and construct validity are interrelated measurement properties. Dimensionality concerns the homogeneity of a scale’s items [49], indicating if the included items match the defined construct. Depression and anxiety have been seen to correlate strongly but are still considered different constructs. Reliability encompasses a scale’s consistency and lack of error [50]. To assess the items' internal consistency, the reliability coefficients of Cronbach’s alpha (α) and composite reliability (ρ_c) were utilized. Finally, construct validity implies various aspects, such as convergent, discriminant, and content validity. In this study, convergent and discriminant validity denote if HADS relates with other constructs as expected, while content validity embraces whether the 14 items adequately represent the theoretical content of the anxiety and depression constructs involved in HADS. Taken together, this is whether the included items cover the theoretical definition they are aimed to represent [51]. If the wording of items is too similar, Cronbach’s alpha, content validity, and dimensionality will be falsely improved. In consequence, the average correlation among items increases, and therefore also coefficient alpha; however, without adding substantially to the content validity of the scale. Obviously, to tap into the same construct, some similarity/correlation is needed. Nevertheless, items simply representing a rephrasing of other items are redundant.

The sample of 1190 adults were aged between 70–96 years with a mean of 76.5 (SD = 5.3); 644 (54.1%) were women (mean age 76.7) while 546 (45.9%) were men (mean age 76.2) (Table 1). Furthermore, 325 (27.3%) had completed higher education (≥ 4 years of university/college). About 50% had a physical or mental long-term illness, injury, or a disorder that impaired their daily functioning (47.1%), only a few had any in-home care (4.7%), in-home nursing care (4.5%) and/or had been admitted to a nursing home for a period during the last year (3.3%).

Looking at those who were excluded, these were significantly older (mean age 80.1; SD = 7.1 years), more often female (60%), and with less education (≥ 4 years of university/college; 9.3%).

The mean anxiety and depression scores were 3.4 (SD = 2.9) and 3.0 (SD = 2.5), respectively (Table 2). The internal consistency of the anxiety and depression constructs (Table 2) was good (α_anxiety = 0.79 =) or acceptable (α_depression = 0.66). Composite reliability (ρc) displayed values between 0.65–0.78 (Table 3); values ≥ 0.60 are acceptable, whereas values ≥ 0.70 are good [43, 47]. Higher symptom scores on HADS correlated significantly with poorer QOL scores (Table 2), supporting convergent validity (H₃, H₄, and H₅).

This study aimed to test the psychometrics of the HADS among community-dwelling people ≥ 70 years. Consequently, first, we tested the original two-factor solution (H₁, H₂, and H₃), including 14 items. This solution was termed Model-1; the factor loadings (λ) ranged between 0.32 and 0.72, followed by multiple squared correlations (R2) from 0.10 to 0.52; this range of factor loadings was pretty much the same in all estimated models. Three items belonging to the depression construct; 8_D (“I feel as I’m slowed down”), 10_D (“I have lost interest in my appearance”) and 14_D (“I can enjoy a good book or a TV program”) revealed low loadings of 0.34, 0.32 and 0.37, respectively, explaining 12%, 11% and 13% of the variance of the depression construct. The fit indices indicated misspecification: χ² = 395.010, p = 0.00001, df = 76, χ²/df = 5.20, RMSEA = 0.059, p-close = 0.04, SRMR = 0.052, CFI = 0.91, TLI = 0.89 (Table 3). The RMSEA, which is an estimate of approximate fit was acceptable, while the χ² was much too high. For an acceptable fit, the χ²/df should be ≤ 3.0, and ≤ 2 for a good fit. Further, the CFI and TLI were too low, all of which indicated misspecification. Exploring the normalized residuals, 23 residuals were significant, with item 6_D (“I feel cheerful “) involved in several highly significant estimates. Hence, we scrutinized the modification indices (MI) presenting some extremely high values; item 6_D exposed an extremely high MI = 77.29 with the anxiety factor and an MI = 30.374 with item 5_A (“Worrying thoughts go through my mind “). Also, item 2_D (“I still enjoy the things I used to enjoy”) and 12_D (“I look forward with enjoyment to things”) demonstrated an exceptionally high MI = 44.383. In total, 16 MIs were ≥ 10.

To further test the original model, we looked at one of the two factors at a time: anxiety and depression, both including 7 items each, demonstrated a too high χ², while the other indices were good to acceptable (Anxiety: χ² = 61.695, p = 0.00001, df = 14, χ²/df = 4.41, RMSEA = 0.054, p-close = 0.320, SRMR = 0.029, CFI = 0.97, TLI = 0.96; Depression: χ² = 84.264, p = 0.00001, df = 14, χ²/df = 6.02, RMSEA = 0.065, p-close = 0.30, SRMR = 0.038, CFI = 0.94, TLI = 0.92) (Table 3). Composite reliability was good, showing estimates of ρ_Anxiety = 0.78 and ρ_Depression = 0.70. Both factors revealed a too-high chi. This often causes high modification indices (MIs), indicating cross-loadings and significant correlations among error terms. Hence, we considered the theoretical content of the items.

It is plausible that if one is feeling cheerful, one is improbable to feel anxiety at the same time, and vice versa. Thus, it is theoretically meaningful that feeling cheerful (item 6_D) and feeling anxious correlate negatively. Accordingly, in Model-2 we included a path (cross-loading) from item 6_D (“I feel cheerful”) to the Anxiety construct, which improved the fit considerably: χ² = 317.951, p = 0.0001, df = 75, χ²/df = 4.24, RMSEA = 0.046, p-close = 0.262, SRMR = 0.042, CFI = 0.93, TLI = 0.92. However, the fit was still not good.

An extremely high MI between the items 2_D (“I still enjoy the things I used to enjoy”) and 12_D (“I look forward with enjoyment to things”) was uncovered. It is rational that still enjoying things and looking forward to things with enjoyment correlate. Therefore, we included a correlated error term between item 2_D and 12_D which further improved the fit in Model-3: χ² = 296.919, p = 0.0001, df = 74, χ²/df = 4.01, RMSEA = 0.050, p-close = 0.452, SRMR = 0.040, CFI = 0.94, TLI = 0.92 (Fig. 1). For Model-3, composite reliability (ρ_c) was 0.78 for the anxiety subscale and 0.65 for the depression subscale. Still, 14 normalized residuals were significant, asking for several cross-loadings and correlated errors.

Therefore, we checked if a unidimensional solution termed Model-4 would fit better, though revealing an exceedingly bad fit: χ² = 840.194, p = 0.0001, df = 77, χ²/df = 10.91, RMSEA = 0.091, p-close = 0.0001, SRMR = 0.070, CFI = 0.78, TLI = 0.74. However, internal consistency was good ρ_HADS = 0.81. Hence, the misspecification was possibly caused by error covariances. Consequently, we turned to Model-3 (Fig. 1), again scrutinizing the MI values. The items 8_D, 12_D and 14_D displayed MIs > 15. These items also displayed low loadings, indicating poor reliability as indicators for the depression construct. Possibly, removing some of these items would improve the model fit. Nonetheless, composite reliability was ρ_Depression = 0.65; hence, removing items would cause a weak construct with even lower reliability.

All tested models revealed a chi-square indicating misspecification. However, as already stated, it is well known that chi-square as a model fit index has limitations. First and foremost, chi-square is sensitive to sample size. The present sample is large (N = 1190). Therefore, we randomly split the data into two equally sized parts (N = 595) termed Sample 1 and Sample 2, representing a sample size more suitable for SEM (structural equation modeling) [36, 47, 52]. We tested the original HADS (Model-1) in both Sample 1 and Sample 2. Except the chi-square showing better values, the fit indices demonstrated a similar pattern as in the total sample: Sample1 χ² = 225.391, p = 0.0001, df = 76, χ²/df = 2.97, RMSEA = 0.058, p-close = 0.074, SRMR = 0.057, CFI = 0.90, TLI = 0.88, ρ_Anxiety = 0.76 and ρ_Depression = 0.67. Sample2 χ² = 257.228, p = 0.0001, df = 76, χ²/df = 3.38, RMSEA = 0.063, p-close = 0.005, SRMR = 0.053, CFI = 0.91, TLI = 0.90, ρ_Anxiety = 0.79 and ρ_Depression = 0.73. Loadings ranged between 0.31 and 0.67 for Sample1 and between 0.29 and 0.77 for Sample2.

Concerning the dimensionality of the HADS, the two-factor model undoubtedly showed the best fit to the present data; the dimensionality of the HADS questionnaire stood out to be unquestionable supporting H₁. The two factors were properly correlated. However, the original two-factor solution did not reveal a good fit. Thus, H₁ was only partly supported.

The second hypothesis (H₂) concerned the reliability of the HADS. All items were significant. Largely, the items revealed good loadings (shown in Fig. 1) accompanied by good multiple-squared correlations (R²) demonstrating good reliability. Nevertheless, particularly three items belonging to the depression construct (8_D,10_D,14_D) demonstrated low factor loadings and, thus poor reliability, explaining very little of the variance in the construct. These three items caused a low reliability coefficient for depression, while anxiety displayed good reliability. Hence, H₂ was not fully supported.

H₃ tested the construct validity, which concerns whether the set of measured items reflects the theoretical latent construct those items are designed to measure. Hence, it deals with the accuracy of measurement involving psychometric evidence of convergent and discriminant validity [58]. A measure is said to process convergent validity if independent measures of the same construct converge or are highly correlated [49]. Usually, researchers do not have data on two different, e.g., depression scales scored by the same sample: this represents a frequent problem connected with convergent validity. However, measures that theoretically are predicted to correlate significantly with depression might be used. The present study included measures of overall global QOL to test for convergent validity, which was supported by a significant correlation in the expected direction.

Testing discriminant validity, H₄ stated that HADS correlates significantly and negatively with QOL, while H₅ expected anxiety and depression to perform as two distinct concepts. Discriminant validity specifically measures whether constructs that theoretically should not be related to each other are, in fact, significantly unrelated. In psychometrics, discriminant validity, also termed divergent validity, indicates that the results obtained by the scale (here HADS) do not correlate too strongly with measurements of a similar but distinct trait; two tests reflecting different constructs should not be strongly related to each other. If they are, we cannot be sure they are not measuring the same construct. Accordingly, discriminant validity indicates the extent of difference between two constructs. The complementary concept to divergent validity is convergent validity; both are forms of construct validity. Hence, a high correlation (higher than 0.40) [59] between HADS and QOL would indicate that the measures substantially overlap and do not behave as clearly distinct constructs [49]. Moreover, a high correlation between anxiety and depression would indicate that the two factors were measuring much of the same trait: this would give a good internal consistency (Cronbach’s alpha and composite reliability) but blur the dimensionality. In this study, the anxiety and depression factors performed like distinct concepts supporting the discriminant validity. Simultaneously, the factor correlation between anxiety and depression was highly significant, supporting convergent validity [49]. The convergent and discriminant validity was further supported by significant correlations in the predicted direction for anxiety and depression towards QOL, supporting hypothesis H₄.

Content validity is a central aspect of construct validity. Reliability and content validity represent interrelated measurement properties. In fact, despite good reliability, content validity might be poor. Contrariwise, validity cannot be good if reliability is low [49]. Item 8_D concerns “I feel as I’m slowed down” demonstrated low reliability and, thereby poor validity. In the present sample, with a mean age of 76.5 years, most individuals outside an active work-life have lots of time to adjust to a slower pace of life. Possibly, ‘feeling slowed down’ does not correspond well to older home-living adults’ daily experiences in relation to depression. This item did not perform to be a valid or reliable indicator of depression in this population. Moreover, about 50% of the participants reported physical or mental long-term illness, injury, or loss of function in daily life. Relevantly, a slower pace of life might seem natural and not necessarily an indicator of depression [7].

Likewise, item 10_D, “I have lost interest in my appearance,” did not communicate well with these older adults, indicating low reliability and content validity. Losing interest in one’s appearance did not act as a valid indicator of depression in this population. Losing interest in one’s appearance may be reasoned by the inevitable age-related changes they experience rather than as a symptom of depression. Moreover, item 14_D, “I can enjoy a good book or TV program,” also stood out as an unreliable indicator of depression. Plausibly, being old, enjoying a good book, or watching TV does not relate to depression. Living in your seventies-eighties-nineties, passive leisure activities are everyday activities that are useful as restoration time after active leisure activities and are related to QOL [60]. Reading books might be more demanding due to a decline in sight as well as fatigue. Consequently, item 14_D did not explain any variance in the depression construct and thus misbehaved as a valid indicator for the depression construct.

These findings are consistent with previous studies among nursing home residents without cognitive impairment [37] and hospitalized older adults [33], where the same three items were troublesome among older adults in Norwegian care facilities. In older ages, for the first time in their life, retired adults can slow down. Also, due to a decline in age-related reserve capacity and fear of falling, the most common fear in older adults [61], many older adults may be forced to a slower pace. Doing passive activities such as watching TV or reading may also be a consequence of having a chronic medical condition and multimorbidity, which is associated with anxiety and depression [62]. Hence, the wording of the items 8_D, 10_D, and 14_D should be carefully considered to improve reliability.

Furthermore, the former validation study among older adults in nursing homes [37] also involved a cross-loading for item 6_D to anxiety, as well as highly significant error variances between items 2_D and 12_D. Surprisingly, community-dwelling older adults living at home (the present study), nursing home residents (two different samples giving an approximate N = 500; mean age 84.5 and 86 years) [37], and hospitalized older adults (N = 484; mean age 80.7 years) [33] respond similarly findings of the HADS-D items.

Summarized, construct validity and reliability of anxiety were good. Conversely, the depression construct revealed low validity and reliability, which are interrelated measurement properties. Exclusively, content validity includes the extent to which elements of a measurement scale are appropriate and characteristic of the specific construct for a certain assessment purpose [49]. In this study, content validity concerns whether the 14 HADS items and the two-factor dimensionality precisely represent anxiety and depression in this population. Besides, evidence of face validity can be considered as one aspect of content validity [49]. High face validity of an instrument increases its use in practical situations via ease of use, proper reading level, clarity, and appropriate response formats. Thus, to improve content validity and thereby also reliability for the depression factor, qualitative studies could be applied to get closer to the actual content of depression, investigating what might be the most essential indicators of depression among community-dwelling older adults. Based on such novel evidence, the three troublesome items could be formulated in a more valid format.

A notable strength of this research is the empirical examination of the HADS, which has not been tested previously in a community-dwelling older population of 70 + using CFA in Norway. Also, the large sample size is a strength, allowing the possibility to randomly split the sample into two different samples, including 595 community-dwelling older adults each.

Although the older adults were selected randomly in two subsamples, we cannot state that the sample represents the community-dwelling older adults in the actual city since 3181 of 4667 declined participation. In addition, those excluded from this present study were older and had less education. Hence, in the view of representativity, we assume that the present sample may be disrupted, not representing all community-dwelling older adults.