Impact of Major Depression and Subsyndromal Symptoms on Quality of Life and Attitudes Toward Aging in an International Sample of Older Adults

Abstract

Purpose: The impact of major depression on quality of life (QOL) and aging experiences in older adults has been reported. Studies have demonstrated that the clinical diagnosis of major depression is the strongest predictor for QOL. We postulate that some findings are biased because of the use of inadequate instruments. Although subsyndromal depression is more prevalent than major depression, there are no reports on its impact on QOL or attitudes toward aging. In the present study we aim at assessing the association of major and subsyndromal depression on QOL and attitudes toward aging in a large international sample. Design and Methods: Our cross-sectional study assessed 4,316 respondents in 20 countries from five continents. The study used the World Health Organization Quality of Life (WHOQOL) Assessment for Older Adults, known as the WHOQOL-OLD; the brief version of the WHOQOL instrument, known as the WHOQOL-BREF; and the Attitudes to Ageing Questionnaire. Statistical analyses involved hierarchical multiple regression, as well as comparison of means.  Results: Even relatively minor levels of depression are associated with a significant decrease in all QOL domains and with a pattern of negative attitudes toward aging (overall WHOQOL-OLD R² change =.421). QOL and attitudes toward aging scores are lower as depression intensity is increased, even in subsyndromal levels (overall WHOQOL-OLD mean scores of 95.7 vs 86.4, p <.001). This phenomenon happens not only for clinically depressed individuals but also for subsyndromic individuals. Implications: Present findings suggest that classifying a respondent as nondepressed is not sufficient and is still not informative about his or her QOL and attitudes toward aging status.

Depression has been recognized as a major mental health problem in older adults. It is also the most prevalent mental condition in elderly individuals (Chan, Chien, Thompson, Chiu, & Lam, 2006). It is currently projected that depression will be the second leading cause of disability worldwide in 2020 (Demyttenaere et al., 2004). Furthermore, it carries a poor prognosis, being associated with increased mortality, morbidity, and the use of health facilities (Covinsky et al., 1999; Dozeman et al., 2007; Rovner, 1993; Street, O'Connor, & Robinson, 2007).

The impact of depression in quantitative outcomes (such as mortality) has been extensively demonstrated in the literature (Covinsky et al., 1999). In addition, the role that major depression plays in the quality of life and aging experiences in older adults has been reported (Chan et al., 2006; Low & Molzahn, 2007). Several studies have demonstrated that the clinical diagnosis of major depression is the most influential predictor for impairments in quality of life (Chan et al.; Netuveli, Wiggins, Hildon, Montgomery, & Blane, 2006; Sobocki et al., 2007; Stafford, Berk, Reddy, & Jackson, 2007), even when confounding factors (such as age, gender, living arrangements, or physical conditions) are controlled.

We postulate that some findings may be biased as a result of the inadequacy of the instruments applied in these investigations. The majority of instruments have been used without adequate validation (Haywood, Garratt, & Fitzpatrick, 2005), do not take into consideration several aspects of life that older adults consider fundamental (Pearlman & Uhlmann, 1988), or are not suitable for older adults because they have been developed primarily for young adult populations (Brazier, Walters, Nicholl, & Kohler, 1996; Chachamovich, Trentini, & Fleck, 2007; Power, Quinn, & Schmidt, 2005).

The prevalence of subsyndromal depression in elderly individuals is more than double the one for major depression (D. G. Blazer, 2003; Geiselmann & Bauer, 2000; Geiselmann, Linden, & Helmchen, 2001; Snowdon, 2001; Watson, Lewis, Kistler, Amick, & Boustani, 2004). In the United States, around 15% of the older adults are expected to present subthreshold depression, which is defined as depressive symptoms that fail to meet the full diagnostic criteria for a major depression episode (Judd, Schettler, & Akiskal, 2002). In contrast to major depression, the prevalence of subsyndromal depression seems to increase with advancing age (D. Blazer & Williams, 1980; Ernst & Angst, 1995; Lavretsky & Kumar, 2003; Tannock & Katona, 1995; VanItallie, 2005). Although the number and severity of symptoms are lower than those in the full-blown syndrome, subsyndromal depression is associated with significant functional impairment and psychosocial disability (Penninx, Leveille, Ferrucci, van Eijk, & Guralnik, 1999; Wilms, Kanowski, & Baltes, 2000), an increased risk of developing major depression at some point, and suicidal ideation (Geiselmann & Bauer; Greden, 2001; VanItallie). However, to our knowledge, there are no reports on the impact that subthreshold depression may have on quality of life or attitudes toward aging.

The brief version of the World Health Organization Quality of Life (WHOQOL) instrument, known as the WHOQOL-BREF instrument, is a generic measurement of quality of life (the WHOQOL Group, 1998). Its development was based on a transcultural simultaneous methodology (Bullinger, Power, Aaronson, Cella, & Anderson, 1996; Guillemin, 1995), which ensures its applicability in different cultural settings. Recent reports have described its satisfactory psychometric properties in older adult populations (Chachamovich et al., 2007; Hwang, Liang, Chiu, & Lin, 2003). In addition, the WHOQOL Assessment for Older Adults (WHOQOL-OLD) and the Attitudes to Ageing Questionnaire (AAQ) instruments have been recently developed to provide more reliable and comprehensive tools in assessing quality of life and attitudes toward aging in older adults (Laidlaw, Power, & Schmidt, 2007; Power et al., 2005). They are in line with the need of broader conceptual outcome measures, beyond the classic ones, such as mortality and morbidity. Because they were developed as specific instruments for elderly individuals, particular attention was paid to guarantee that the specificities of old age would be properly covered by the instruments. Both instruments were developed with the use of a combination of classical and modern psychometric approaches, which ensures that the instruments have satisfactory psychometric performance (Laidlaw et al.; Power et al.). The WHOQOL Group recommends that the WHOQOL-OLD module be applied together with one generic quality of life instrument (WHOQOL-100 or WHOQOL-BREF) to enhance its validity (Power et al.).

Our aim in this study is to assess the association of depression symptoms on the quality of life and attitudes toward aging in a large international sample of older adults. Furthermore, we explore to what extent subclinical depression symptoms also determine alterations in the quality of life and attitudes toward aging in this population.

Methods

Participants

We obtained data from a larger study on quality of life in older adults (WHOQOL-OLD). This project was carried out by the WHOQOL Group and involved 20 countries around the world (Power et al., 2005).

Researchers recruited opportunistic samples of individuals in participant centers across Europe, Asia, South America, and North America. A total of 4,316 adults aged 60 years old or older were interviewed and completed the 15-item Geriatric Depression Scale (GDS), WHOQOL-OLD, AAQ, and a sociodemographic form. Researchers carried out the recruitment in university hospitals, nursing homes, and in community groups for older adults.

Procedures

The assessment was run in a cross-sectional design. As we already mentioned, the participants were required to complete a sociodemographic form, the 15-item GDS (Sheikh & Yesavage, 1986) the WHOQOL-BREF instrument (the WHOQOL Group, 1998), the WHOQOL-OLD module (Power et al., 2005), and the AAQ (Laidlaw et al., 2007).

The GDS is a widely used instrument for assessing depression specifically in older adults. Its theoretical background is based on the ability to distinguish depressed elders from the ones with “normal aging” neurovegetative symptoms. Thus, the GDS focuses on nonsomatic symptomatology, emphasizing the affective and cognitive areas (de Craen, Heeren, & Gussekloo, 2003; McDonald et al., 2006; Yesavage et al., 1982). It has yet to be cross-nationally validated (this issue is currently being addressed elsewhere). Thus, each center has applied a nationally validated version. The optimal cutoff score of 5/6 presents satisfactory sensitivity and specificity (Almeida & Almeida, 1999; Rinaldi et al., 2003; Sutcliffe et al., 2000; Wancata, Alexandrowicz, Marquart, Weiss, & Friedrich, 2006), although alternative cutoff points have been suggested for specific conditions (Cullum, Tucker, Todd, & Brayne, 2006; Lim et al., 2000; Malakouti, Fatollahi, Mirabzadeh, Salavati, & Zandi, 2006). Because the instrument has yet to be internationally validated, its cutoff point is not consensual. For the present study, we adopted the 5/6 cutoff point.

The WHOQOL-BREF instrument is a generic questionnaire to assess quality of life. Developed under a simultaneous transcultural approach (Bullinger et al., 1996; Guillemin, 1995), it is suitable for multicentric studies and is also validated for older adults (Chachamovich et al., 2007; Hwang et al., 2003; Naumann & Byrne, 2004). It is composed of four domains (physical, psychological, social, and environmental) and an overall score.

The WHOQOL-OLD module is a specific complementary set of items recently developed to increase the specificity of the quality of life measurement in older adults (Power et al., 2005). It comprises six domains (sensory abilities, autonomy, past, present, and future activities, social participation, death and dying, and intimacy) and an overall score. Furthermore, it proved to be suitable and adequate to measuring quality of life comprehensively in older adults, because an intense qualitative phase ensured that relevant issues were properly covered by this instrument (Chachamovich et al., 2007; Power et al.).

The AAQ is an instrument developed by the WHOQOL Group to assess the subjective perception of aging. It also followed a transcultural methodology and shows good psychometric performance (Laidlaw et al., 2007). It is composed of three facets (psychological growth, physical change, and psychosocial loss) and was also based in an intense theoretical debate among international experts, as well as in focus groups carried out with older adults to confirm or adjust the instrument items. It is also able to assess attitudes toward aging from the perspective of the elderly population, which is the one researchers consider to be the most capable of informing them about the experience of aging.

Statistical Analysis

We ran our statistical analyses by using SPSS 14.0 software. We analyzed the total sample through descriptive statistics for both clinical and subsyndromal depression. If we found the distribution of demographic variables to be significantly different, then we ran an analysis of covariance (ANCOVA) to partial out the variance for which other predictors (age, educational level, gender, and marital status) could account (Field, 2005). We carried out independent t tests to compare the quality of life and attitudes toward aging scores according to the predictors.

We applied hierarchical regression in each domain of the WHOQOL-OLD, WHOQOL-BREF, and AAQ instruments to detect the impact of depression. We included age, gender, educational level, and marital status as independent variables, together with depression levels.

Next, we selected the subsample of individuals with subsyndromal depression levels and divided them into two groups (one with GDS scores from 0 to 2, and another with scores from 3 to 5). We based this cutoff on the median of the GDS score distribution. Again, we applied demographic distribution verification and ANCOVA to ensure comparability. We ran independent t tests for each domain. At this stage, these comparisons had the aim of detecting the impact of small increases in depression scores even at a subsyndromic level. We also carried out a hierarchical regression to detect the relative weight of subsyndromic depression on quality of life and attitudes toward aging.

Results are described through means and probability or p values (chi-square and t tests), and squared multiple correlation or R² changes and standardized beta or β values (for hierarchical multiple linear regression).

Results

Descriptives

Researchers assessed 4,316 respondents. The sample was predominantly comprised of highly educated married women. Researchers carried out recruitment primarily in homes (76.6%), followed by community sites (14.7%) and hospitals (8.7%). We found no relevant correlation between the site of recruitment and the domain scores for quality of life or attitudes toward aging (Pearson coefficients from 0.065 to 0.155). Table 1 describes the demographic characteristics of the total sample divided into a subsyndromal group and a clinical depression group. There is a marked predominance of respondents with depression scores lower than the diagnostic cutoff point (73.84%).

We compared the WHOQOL-BREF, WHOQOL-OLD, and AAQ domain scores across the clinically depressed and the subsyndromic groups. All scores for quality of life and attitudes toward aging are significantly lower in the depressed group than in the subsyndromic group (p <.001; data not shown). These findings point out a wide and intense association of major depression in quality of life outcomes.

This association, however, could be partially related to the discrepancy in the distribution of demographics. To control this potential bias, we ran an ANCOVA. Table 2 describes the ANCOVA results. As shown in this table, the interaction among these variables assumes statistical significance in only 4 out of the 16 scores. Furthermore, the interaction explains not more than 3% of the total variance of the score in these four domains. Thus, the difference in scores for quality of life and attitudes toward aging between the two samples appears to be strongly related to depression levels, and not to demographic dissimilarities.

We tested potential multicollinearity through the variance inflation factor in each regression model. Variance inflation factor values ranged from 0.903 to 1.09, which indicates that the predictors do not have strong linear relationships among them (Field, 2005; Hair, Tatham, & Black, 1998).

We conducted a hierarchical multiple linear regression to detect the increase of the coefficient of determination as new independent variables were included in the model. We evaluated the data distribution by means of the Kolmogorov–Smirnov test, which showed a normal distribution of all dependent variables (WHOQOL and AAQ domains scores). We also checked linearity and variance and these met the linear regression assumptions. We selected gender, age, marital status, and educational level to compose the regression model, together with depression levels, because they represent important demographic variables and are widely reported in several studies (Chan et al., 2006; Low & Molzahn, 2007). We also examined the standardized beta coefficients to compare the impact of the independent variables in the quality of life and attitudes toward aging domains. Tables 3 and 4 illustrate the R² increase and the beta coefficients for each domain, as well as for the overall scores in the three scales.

The inclusion of depression in the model determined a marked increase in the coefficient of determination. In fact, almost all of the variance is explained by depression. The magnitude of the effect caused by depression in the model is greatly higher than the ones by the other variables, as we can see by both the R² change and the beta coefficients.

Interestingly, the domains of death and dying and psychological growth seem to be less related to depression. In fact, the proposed model explains only 8.4% and 7.3% of the total variance, respectively. As we expected, depression is negatively associated with all scores in different intensities.

Besides testing the association of major depression on quality of life and attitudes toward aging, we also explored the subsyndromal symptomatology. Our main objective in these further analyses was to investigate whether an increase of depression levels can be related to impairments in quality of life and attitudes toward aging, even when they are not considered clinically relevant.

We selected a subsample of 3,187 respondents from the total sample. We included all respondents with GDS scores of 5 or lower in this analysis stage. To test the effect of subsyndromal depression in the WHOQOL and AAQ scores, we divided the sample into two groups. The first group was composed of 1,782 respondents with GDS scores equal to or lower than 2. The second group included respondents with GDS scores from 3 to 5. Demographics are described in Table 5.

We tested demographic variables through the use of chi-square and independent t test statistics. Following the same strategy applied for the total sample and as already described, we used an ANCOVA test to assess a potential effect of this different distribution. Results indicated that the interaction among gender, educational level, and marital status is not significant for any domain. Partial eta-squared values were 0% for all 16 scores (data not shown).

Table 6 shows the comparison of mean scores for all domains. It is possible to observe that the quality of life and attitudes toward aging scores are significantly decreased as depression levels are increased.

In fact, the analysis of the scores of all domains indicates a progressive impairment with increasing depression levels, even when nonclinical depression is included. Figure 1<--CO?1--> illustrates the decreasing quality of life and AAQ overall scores in four different levels of depressive symptoms. An analysis of variance showed that the differences are statistically significant at all points (p <.0001). In addition, it seems that the AAQ scores are less intensely associated with the increase of depression levels when compared with the quality of life scores.

We also applied a hierarchical multiple linear regression in this subsample. Again, we assessed the increase in the coefficient of determination, as well as the standardized beta coefficients. As we expected, the variances of the domain scores were lower than the one observed in the total sample. As a result, total R² and changes in the coefficient were smaller.

The results indicate the same pattern observed in the findings from the total sample. Again, the magnitude of the effect caused by the inclusion of depression in the model is much higher than the ones caused by the other variables. This suggests that a slight increase in depression level (in a subsyndromal scenario) is a major predictor of impairments in quality of life and attitudes toward aging.

The psychological growth and the death and dying domains again showed a low coefficient of determination. However, most of this low value was due to depression rather than any other independent variable in the model. Tables 7 and 8 describe the hierarchical regression findings.

Discussion

In this study we aimed at exploring the association of clinical and subclinical depression in the quality of life and attitudes toward aging in a large international sample of older adults. The results suggest that even relatively minor levels of depression are associated with a significant decrease in quality of life and with a pattern of negative attitudes toward aging. These findings assume particular relevance because the instruments that were used are indeed adequate for investigations in older adults. This ensures the reliability of the results. Furthermore, the large international sample could be seen as an important factor of external validity.

Studies regarding attitudes toward aging have demonstrated the effect of culture in the perception of aging. Interestingly, stereotypes of aging proved not to be adequate to describe the aging process (Yun & Lachman, 2006). Quality of life is also intimately linked to cultural aspects, because culture and value systems of the society are part of the quality of life definition (Schmidt & Bullinger, 2003; the WHOQOL Group, 1998). The AAQ and the WHOQOL instruments were constructed under a simultaneous methodology, which included input from several cultural contexts in the development phase. Thus, they seem to represent what Draper calls a “culture-proof variable,” because they are able to assess meaningfully abstract concepts (Draper, 2007).

Researchers have yet to explore the link between quality of life and attitudes toward aging. Some similarities should be highlighted. First, both concepts are predominately based on subjective perception rather than objective conditions. Second, both are multidimensional and include physical and psychological aspects. Nevertheless, it is possible to state that the AAQ assesses a more stable perception, whereas the WHOQOL instrument would be related to a more circumstantial aspect. The combination of the WHOQOL-OLD and the AAQ is suggested as an effective methodology for assessing psychological or psychiatric interventions (Laidlaw et al., 2007).

The sequential comparisons of the mean scores of quality of life and attitudes toward aging domains indicate progressive impairment on both outcomes with increasing depression levels. This finding is clinically relevant and is in line with the findings from a series of recent studies (Chan et al., 2006; Netuveli et al., 2006; Sobocki et al., 2007; Stafford et al., 2007). However, to our knowledge, the consistent result of impairment in all quality of life and AAQ domains with a slight increase in depression levels (in a subsyndromal stage) has not yet been reported. When analyzed as a whole, the present results show that there is a decrease in quality of life and AAQ as depression levels increase. This phenomenon happens not only for the clinically depressed respondents (i.e., respondents with 6 to 9 symptoms and respondents with 10 or more symptoms on the GDS) but also for subsyndromic respondents.

Thus, findings suggest that classifying a person as nondepressed is not sufficient and is still not informative about his or her quality of life and attitudes toward aging status. Historically, depression has been underdiagnosed and undertreated in later life (Alexopoulos, 2005), and this is evidently more likely to occur if a person does not meet diagnostic criteria (Jeste, Blazer, & First, 2005). These results would raise the hypothesis that subsyndromal depression could be treated because of the impact of minor scores on the GDS on appraisals of aging and quality of life. The potential increase in the quality of life and attitudes toward aging of patients treated for subsyndromic depression symptomatology appears as a question to be further addressed in longitudinal studies.

The hierarchical multiple regressions produced some interesting results. First, depression accounted for the vast majority of coefficients of determination in all domains. The standardized beta coefficients also showed a high difference of magnitude between depression and age, gender, marital status, and educational level. Furthermore, depression was the only predictor with significant values for all outcomes in both total and subsyndromal samples. In fact, depression has been described as the predominant factor contributing to morale (Woo, Ho, & Wong, 2005), as well as the strongest predictor for subjective perception of quality of life (Chan et al., 2006; Demura & Sato, 2003). The present findings corroborate these previously reported results.

Second, there is a marked difference in the extent to which the regression model explained the scores variances. It is suggested that moderate coefficients of determination are expected for models including quality of life because it is a wide and comprehensive construct, which would require complex models to provide adequate explanation (Bowling, Banister, Sutton, Evans, & Windsor, 2002). For example, recent studies using linear regression methodology on quality of life reported model R² values of.568 (Low & Molzahn, 2007) and.475 (Netuveli et al., 2006). Because the model applied in the present study is simpler than the ones just cited, we would expect the R² values to be lower. In fact, the values in the present study ranged from.214 to.476 (with the exception of death and dying and psychological growth) in the total sample. These values are considered sizeable given the complex nature of the dependent variables (Bowling et al.).

In contrast, the coefficient of determination was markedly smaller for the death and dying and psychological growth domains, which suggests that there may be other important and specific factors that account for the variation in these areas. Our data were not able to address this question, though. It is important to note that the death and dying domain seems to present some potential psychometric weaknesses, as demonstrated by other reports (Chachamovich et al., 2007; Power et al., 2005). We hypothesize that this domain would perform better when it is applied to unhealthy subjects but would lose power when healthy subjects are included in the sample. Studies on quality of life in HIV patients have supported this hypothesis (O'Connell, Skevington, & Saxena, 2003; O'Connell, Saxena, & Skevington, 2004).

Regarding the psychological growth domain, the only article published up to the present that we know of describes its satisfactory psychometric performance in an international data set (Laidlaw et al., 2007), which indicates that there must be another cluster of variables for determining psychological growth, and that these variables are more specific for this domain. A close look at the eight items included in this domain suggests that generativity is one major topic (i.e., passing on the benefits of growing older, giving good examples to younger people; see Laidlaw et al.). It is possible to hypothesize that older adults without the opportunity to be in contact with younger generations (e.g., without children or grandchildren) may not be adequately represented by this domain. It is extremely relevant to design investigations on this topic in order to implement specific interventions to improve psychological aspects of growing older.

As we predicted, the same model produced lower coefficients of determination in the subsyndromal sample. However, the results shown for the total sample (i.e., elevated and significant standardized beta coefficients for depression in all domains and the pattern of R² values) remained similar for the subsyndromic sample. This finding reinforces the role of even minimal depression levels in the WHOQOL and AAQ domain scores. The question that arises from these findings would be whether offering treatment for subsyndromal patients (either psychopharmacological or psychotherapy) could improve their quality of life and attitudes toward aging. Studies designed specifically to address this issue are needed.

There are some limitations in the present study. The cross-sectional design does not allow inference of causality. Thus, it is possible to detect an association between depression and WHOQOL and AAQ scores, but not to state that the latter is caused by the former. Longitudinal studies are required to address this issue. In addition, clinical or structured psychiatric interviews were not used to provide a clinical diagnosis of depression. However, the GDS was chosen because it has been extensively studied for diagnostic purposes. Furthermore, we decided to use level of depression rather than the clinical diagnosis of depression in the analysis to avoid this potential bias. It is also important to observe that the tested model did not include other demographic, clinical, and social variables, which could be potentially related to depression and quality of life (such as impaired functioning, life stressors, social support, and physical health). These nonevaluated variables may play a role in the relationship between depression levels and quality of life.

In summary, the present study showed that an increase in levels of depression, both in clinical and subsyndromic stages, is a major predictor of impairments in quality of life and attitudes toward aging for older adults.

References