- Email: [email protected]
The Built Environment and Depression in Later Life: The Health In Men Study
The Built Environment and Depression in Later Life: The Health In Men Study Dick Saarloos, Ph.D., Helman Alfonso, Ph.D., Billie Giles-Corti, Ph.D., Nick Middleton, B.Sc. Hons., Osvaldo P. Almeida, M.D., Ph.D., F.R.A.N.Z.C.P., F.F.P.O.A.
Objective: This study examined the impact of built environment (BE) attributes on depression in older men to determine whether associations were independent of neighborhood composition factors and sociodemographic, psychosocial, and health factors at the individual level. Methods: The authors used geocoded data from the Health in Men Study collected in Western Australia in 2001 (N = 5,218). Depression was measured using the self-rated 15-item Geriatric Depression Scale. Geographic Information Systems were used to objectively measure BE attributes. Univariate logistic regressions were applied to select relevant covariates. Multivariate logistic regressions were conducted to examine BE attributes both separately and conjointly. Results: Higher degrees of land-use mix were associated with higher odds of depression independent of other factors, including street connectivity and residential density (odds ratio = 1.54, 95% confidence interval [CI] = 1.10–2.16, and odds ratio = 1.52, 95% CI = 1.08–2.14 for the second and third tertiles, respectively). Further examination showed that retail availability was associated with a 40% increase in the odds of depression (95% CI = 4%–90%) independent of other factors, including availability of other land uses. Conclusions: The BE is independently associated with depression through land-use mix, and specifically through retail availability. Although local retail facilitates walking, our findings suggest that it may increase the odds of depression in older men. This requires further exploration but suggests the need for careful planning of retail in residential environments, particularly near housing for older adults. (Am J Geriatr Psychiatry 2011; 19:461–470) Key Words: Ageing, elderly, men, depression, depressive disorder, mood disorder, mental health, risk factors, built environment, social context, epidemiology
D
epression is the most common and disabling functional disorder among ageing men.1 In Australia, clinically significant depression affects about ∼8.6% of older men.2 Many factors have
been associated with late-life depression.3 Besides individual characteristics, it has been suggested that social, behavioral, and environmental factors play a role in predisposing, precipitating,
Received November 23, 2009; revised April 19, 2010; accepted April 20, 2010. From the Centre for the Built Environment and Health, School of Population Health (DS, BG-C, NM); Western Australian Centre for Health and Ageing, Centre for Medical Research (HA, OPA); School of Psychiatry and Clinical Neurosciences (HA, OPA), The University of Western Australia, Australia; and Department of Psychiatry, Royal Perth Hospital, Australia (OPA). Send correspondence and reprint requests to Osvaldo P. Almeida, M.D., Ph.D., F.R.A.N.Z.C.P., F.F.P.O.A., WA Centre for Health and Ageing (M573), The University of Western Australia, 35 Stirling Highway, Crawley, WA 6009, Australia. e-mail: [email protected] c 2011 American Association for Geriatric Psychiatry DOI: 10.1097/JGP.0b013e3181e9b9bf
Am J Geriatr Psychiatry 19:5, May 2011
461
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Built Environment and Depression in Later Life and perpetuating depressive symptoms.3–6 Although much attention has been given to the influences of impaired social support3,5 and unhealthy behaviors,7–9 insight into the contribution of environmental factors remains limited. Hitherto, most studies have generally found that depression is less common in neighborhoods of higher socioeconomic status and more common in areas with greater social disorder.10 Other studies, however, have questioned the significance of these relationships.11,12 This inconsistency may well be due to a lack of attempts to examine the physical characteristics of the environment.10 It is argued that the built (or physical) environment becomes more important in later life.13–15 When people age, their life sphere often contracts to their residential environment due to changes such as physical decline, retirement, or decreased access to transport.16–18 This may also shrink their social networks.17 Hence, especially older people’s health and well-being may be affected by the physical and social settings of their neighborhoods. As such, some authors have suggested that the built environment (BE) may contribute to late-life depression.19 In a pioneering study, Berke et al.20 reported that depression was less prevalent among older men living in more walkable neighborhoods. The use of a composite measure such as “walkability,” however, fails to reveal the specific features of the BE that matter most for depression. To facilitate translation into practice,21 it will be necessary to investigate BE attributes that are specific and modifiable. Furthermore, it seems imperative to do so while accounting for social factors. Numerous studies have stressed the importance of social support and sense of community as buffers to stressors that lead to depression, both in the overall population and in older people.5,22,23 To our knowledge, it has not yet been determined to what extent BE attributes are significantly associated with depression above and beyond social factors. We conducted this study to examine the impact of BE attributes on depression in older men, taking an ecological approach that accounts for compositional neighborhood factors and sociodemographic, psychosocial, and health factors at the individual level. The study was designed to ascertain whether associations between BE attributes and depression in older men are independent of these other factors.
462
METHODS Sample and Study Design The study utilized data collected in the Health in Men Study (HIMS). Full details of the study design have been published elsewhere.24 Briefly, HIMS arose out of a population-based randomized trial that targeted men age 65–79 years living in Perth metropolitan area (Western Australia) in 1996–1999. All men were identified based on the electoral roll (enrolment to vote being compulsory in Australia). A followup survey was undertaken in 2001–2004. Our sample was taken from this follow-up survey and consisted of 5,218 older men whose residential addresses were geocoded. Data Preparation Geographic information systems were used to measure BE attributes for each Census Collection District (CCD) with one or more participants. The Western Australia Department for Planning and Infrastructure supplied land-use point data from the Western Australian Land Information Authority. Cadastral data were used to identify all land parcels in residential use. Parcels that overlapped CCD boundaries were intersected, and only the area within a CCD was counted. The Australian Bureau of Statistics provided data on the number of dwellings at the CCD level. A process was developed in which, for each cadastral parcel, the main land use was determined given the land-use classes defined by Department for Planning and Infrastructure and based on rating assessments of the Valuer Generals Office and Reserve Reports from Landgate. For the purpose of this study, where parcels had multiple land uses, we selected the main land use in the following order of preference: “retail” (= activities involving the sale of goods from shops located separate to and/or in a shopping centre), “other retail” (= activities that by virtue of the scale and special nature of goods differ from regular shops, e.g., car sales yard and carpet showroom), “offices and business,” “welfare, health and community services” (= activities providing the community with a specific service, such as hospitals, schools, and religious activities), and “entertainment, recreation, culture.”
Am J Geriatr Psychiatry 19:5, May 2011
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Saarloos et al Measures The prevalence of depression was assessed using the self-rated 15-item Geriatric Depression Scale.25 Participants with a total score of 7 or more were considered to have clinically significant depressive symptoms.26 Several modifiable BE attributes were assessed according to current knowledge on environmenthealth relationships.27–29 We measured street connectivity, residential density, and land-use mix in each person’s neighborhood. Street connectivity is an indicator of the interconnectedness of the street network in an area.28,30 We measured the number of intersections (nodes where three or more street segments meet) in a CCD divided by the area of that CCD (in km2 ). Residential density indicates the average density of residential developments in an area.30,31 This was measured as the total number of dwellings in a CCD divided by the total area of parcels in residential use within that CCD (in hectares). Land-use mix is an indicator of the diversity of land uses in an area.28 We assessed the five land-use classes mentioned and created an index with values ranging from zero (i.e., none of the land uses present) to one (i.e., an even mix of all five land uses). To evaluate neighborhood walkability, we standardized and summed the three measures for each area. Finally, we created measures indicating the availability of each land use in a CCD. Four categories of covariates were ascertained. First, factors of neighborhood composition were derived from census data.10 We obtained an index of relative socioeconomic disadvantage from the Socioeconomic Index for Areas 2001 dataset.32 This index takes into account area attributes such as low income, low educational attainment, unemployment, unskilled jobs, and variables that reflect disadvantage (e.g., indigenous and divorced). The higher a CCDs score on this index, the less disadvantaged it is compared with other areas. To investigate influences of age composition,33,34 we identified the percentages of children (0–14 years), younger adults (15– 24 years), and older adults (65 years and older) in each CCD from census data.35 Second, several individual sociodemographic characteristics were identified.10,36 We retrieved information from participants about their age, place of birth, and education level. We also asked them about their living arrange-
Am J Geriatr Psychiatry 19:5, May 2011
ments (With whom do you live now? [alone/with spouse only/with other relatives/with other nonrelatives/nursing home or hostel/other]), and their type of housing (In what kind of building do you live? [own home/flat, unit/elderly unit/nursing home, or hostel/other]). Third, several psychosocial factors were measured.5,37 We used the Duke Social Support Index (DSSI) to measure people’s satisfaction with their network of relationships. The DSSI generates scores ranging from 6 to 18, with higher scores indicating greater satisfaction. The Family and Friends Adaptation, Partnership, Growth, Affection, Resolve (APGAR) Scale38 was used to assess how participants judge the functioning of their social network, where higher scores suggest better functioning. Also, a 13-item measure of sense of community was included.39 The fourth category of covariates involved individual health factors. Physical comorbidity was measured with the weighted Charlson Comorbidity Index,40 which takes into account both number and seriousness of 17 common medical conditions that predict 1-year mortality. Comorbidity data were retrieved from the Western Australia Data Linkage System using established coding algorithms.41 Also, participants were asked about their smoking behavior (Have you ever smoked regularly? [yes/no]; if yes, how often do you smoke now? [every day/not every day/not at all]) at the time of assessment.42 Statistical Analysis BE attributes were transformed into categorical variables guided by the distribution of observed values. The five land-use availability attributes were measured as binary variables, indicating whether or not any part of a participant’s CCD was occupied by a specific land use. Land-use mix was converted into tertiles, whereas walkability, street connectivity, and residential density were split into quartiles. All covariates measuring neighborhood composition were turned into tertiles. Individual sociodemographic covariates were categorized into groups with sufficient observations. In case of DSSI, we allocated participants to three groups: 6–14 (lowest), 15–17 (middle), and 18 (highest). The APGAR Scale and sense of community measure were transformed into tertiles. Comorbidity and smoking behavior were operationalized into legitimate categories.
463
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Built Environment and Depression in Later Life We used Stata 10.1 software to analyze the data.43 Alpha was set at 5%, and all tests performed were two-tailed. The conceptual framework underpinning the analyses was based on the assumption that characteristics of the BE that facilitate social interactions decrease the risk of depression and that demographic, health, and psychosocial factors may confound this association. Therefore, we completed the analyses in three steps. First, univariate logistic regressions were conducted to assess the crude associations of walkability and its three components (street connectivity, residential density, and land-use mix) with depression and to select covariates significantly associated with depression. Second, multivariate logistic regressions were applied to further examine the BE attributes both separately and conjointly, while adjusting for selected covariates. Third, we unpacked the land-use mix variable to investigate the role of specific land uses by means of multivariate logistic regressions.
RESULTS A total of 295 men (5.7%) reported clinically significant depressive symptoms. Table 1 shows the univariate associations of all BE attributes and covariates with depression. Among the BE attributes, residential density was associated with depression, because those in the highest quartile were more likely to be depressed (odds ratio [OR] = 1.44; [95% confidence interval {CI} = 1.03–2.02]). The same trend was found for land-use mix, with significantly higher odds found for the highest tertile (OR = 1.37 [95% CI = 1.02–1.84]). Among factors of neighborhood composition, the Socioeconomic Index for Areas disadvantage index showed that the odds of depression were significantly lower in the least disadvantaged areas (OR = 0.69 [95% CI = 0.51–0.92]). Regarding individual sociodemographics, higher odds of depression were found among older men, those born overseas, those living alone, and those not living in their own home. Conversely, the odds of depression were lower in men who had an education level of at least high school. All included psychosocial factors revealed inverse relationships with depression. The odds were significantly lower in the highest tertiles of DSSI (OR = 0.05 [95% CI = 0.03–0.07]), APGAR (OR = 0.19 [95% CI = 0.13–0.28]), and sense of community
464
(OR = 0.34 [95% CI = 0.25–0.47]). With respect to health factors, we found higher odds of depression related to higher comorbidity and smoking in past and present. Next, we investigated the association between depression and the three components that make up the walkability index, controlling for covariates that showed a significant univariate association with depression (Table 2). Only land-use mix was significantly associated with depression independent of neighborhood composition and individual level factors. After additionally controlling for the effects of street connectivity and residential density (Model 2e), older men living in areas with the two highest tertiles of land-use mix were on average 53% more likely to be depressed than those in the lowest tertile of land-use diversity. When examined, 89% of areas in the lowest tertile contained none of the land uses, whereas 97% of areas in the middle tertile had one or two land uses present, and all areas in the highest tertile included two or more land uses. Compared with the covariates, the effect of land-use mix on depression was similar in size to the effects of educational attainment and APGAR. The strongest was the effect of DSSI with the odds of depression being 11 times lower in those being most satisfied (OR = 0.09 [95% CI = 0.05–0.14]). At an intermediate level, age, comorbidity, and smoking behavior were related to increased odds of depression. Finally, the land-use mix indicator was dissected into five land-use availability variables to examine the effects of specific land useson in older men with retail in their CCD were higher (OR = 1.46 [95% CI = 1.11–1.90]) than in areas without retail. There was no interaction between land-use mix and retail availability. When additionally controlling for other landuse availability variables (Model 3f), the effect of retail availability remained significant (OR = 1.40 [95% CI = 1.04–1.90]). We also examined the availability of parks and gardens, a subset of “entertainment/ recreation/culture,” but its effect on depression was insignificant. Reverse causality may exist in the sense that people more prone to depression may choose to live in neighborhoods with mixed land use or with retail. To assess whether this “social drift”10,19,44 may have occurred in our sample, we compared the results for participants with varying degrees of depression (“no depression” [Geriatric Depression Scale score =
Am J Geriatr Psychiatry 19:5, May 2011
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Saarloos et al
TABLE 1.
Univariate Associations With Depression (Significant Effects Highlighted in Bold)
Variables Built environment Walkability
Street connectivity
Residential density
Land-use mix
Neighborhood composition SEIFA disadvantagea
Percentage of children (0–14 years)
Percentage of younger adults (15–24 years)
Percentage of older adults (65+ years)
Individual sociodemographics Age group (years)
High school education Migrant Living alone Housing type
Individual psychosocial factors DSSI (social support)
APGAR (family and friends)
Sense of community
Individual health Charlson comorbidity
Smoking
Total
Depression (n = 295)
No Depression (n = 4,923)
Odds Ratio (95% CI)
Q1 (low) Q2 Q3 Q4 (high) Q1 (low) Q2 Q3 Q4 (high) Q1 (low) Q2 Q3 Q4 (high) T1 (low) T2 T3 (high)
1,317 1,302 1,308 1,291 1,340 1,282 1,352 1,244 1,314 1,314 1,302 1,288 1,749 1,744 1,725
63 (21.4) 78 (26.4) 75 (25.4) 79 (26.8) 81 (27.5) 68 (23.1) 78 (26.4) 68 (23.1) 62 (21.0) 65 (22.0) 82 (27.8) 86 (29.2) 82 (27.8) 104 (35.3) 109 (36.9)
1,254 (25.5) 1,224 (24.9) 1,233 (25.0) 1,212 (24.6) 1,259 (25.6) 1,214 (24.7) 1,274 (25.9) 1,176 (23.9) 1,252 (25.4) 1,249 (25.4) 1,220 (24.8) 1,202 (24.4) 1,667 (33.9) 1,640 (33.3) 1,616 (32.8)
1 1.27 (0.90–1.78) 1.21 (0.86–1.71) 1.30 (0.92–1.82) 1 0.87 (0.62–1.21) 0.95 (0.69–1.31) 0.90 (0.64–1.25) 1 1.05 (0.74–1.50) 1.36 (0.97–1.91) 1.44 (1.03–2.02) 1 1.29 (0.96–1.74) 1.37 (1.02–1.84)
T1 (low) T2 T3 (high) T1 (low) T2 T3 (high) T1 (low) T2 T3 (high) T1 (low) T2 T3 (high)
1,713 1,707 1,748 1,731 1,742 1,734 1,754 1,713 1,740 1,746 1,727 1,734
116 (39.3) 94 (31.9) 83 (28.1) 103 (34.9) 95 (32.2) 97 (32.9) 110 (37.3) 94 (31.9) 91 (30.8) 100 (33.9) 84 (28.5) 111 (37.6)
1,597 (32.4) 1,613 (32.8) 1,665 (33.8) 1,628 (33.1) 1,647 (33.5) 1,637 (33.3) 1,644 (33.4) 1,619 (32.9) 1,649 (33.5) 1,646 (33.4) 1,643 (33.4) 1,623 (33.0)
1 0.80 (0.61–1.06) 0.69 (0.51–0.92) 1 0.91 (0.68–1.21) 0.94 (0.70–1.25) 1 0.87 (0.65–1.15) 0.82 (0.62–1.10) 1 0.84 (0.62–1.13) 1.13 (0.85–1.49)
69–74 75–79 80–84 85+ No Yes No Yes No Yes Own home Flat/unit Other
2,465 1,752 821 160 2,724 2,252 2,972 2,006 4,355 843 4,099 709 406
106 (35.9) 114 (38.6) 57 (19.3) 17 (5.8) 185 (62.7) 98 (33.2) 148 (50.2) 135 (45.8) 232 (78.6) 62 (21.0) 210 (71.2) 52 (17.6) 32 (10.8)
2,359 (47.9) 1,638 (33.3) 764 (15.5) 143 (2.9) 2,539 (51.6) 2,154 (43.8) 2,824 (57.4) 1,871 (38.0) 4,123 (83.7) 781 (15.9) 3,889 (79.0) 657 (13.3) 374 (7.6)
1 1.55 (1.18–2.03) 1.66 (1.19–2.31) 2.65 (1.54–4.54) 1 0.62 (0.49–0.80) 1 1.38 (1.08–1.75) 1 1.41 (1.06–1.89) 1 1.47 (1.07–2.01) 1.58 (1.08–2.33)
Low Medium High T1 (low) T2 T3 (high) T1 (low) T2 T3 (high)
875 2,087 2,226 1,823 1,847 1,482 1,728 1,771 1,672
166 (56.3) 100 (33.9) 25 (8.5) 196 (66.4) 57 (19.3) 33 (11.2) 150 (50.8) 86 (29.2) 53 (18.0)
709 (14.4) 1,987 (40.4) 2,201 (44.7) 1,627 (33.0) 1,790 (36.4) 1,449 (29.4) 1,578 (32.1) 1,685 (34.2) 1,619 (32.9)
1 0.21 (0.17–0.28) 0.05 (0.03–0.07) 1 0.26 (0.20–0.36) 0.19 (0.13–0.28) 1 0.54 (0.41–0.71) 0.34 (0.25–0.47)
0 1–2 3–4 5+ Never Past Current
3,807 686 484 241 1,703 3,228 264
161 (54.6) 48 (16.3) 52 (17.6) 34 (11.5) 59 (20.0) 209 (70.8) 26 (8.8)
3,646 (74.1) 638 (13.0) 432 (8.8) 207 (4.2) 1,644 (33.4) 3,019 (61.3) 238 (4.8)
1 1.70 (1.22–2.38) 2.73 (1.96–3.78) 3.72 (2.50–5.52) 1 1.93 (1.44–2.59) 3.04 (1.88–4.92)
Level
Notes: Data are given as N (%) or OR with 95% CIs. Q: Quartile; T: Tertile. SEIFA: Socioeconomic Index for Areas. a Higher values on the SEIFA disadvantage index imply less disadvantage.
Am J Geriatr Psychiatry 19:5, May 2011
465
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Built Environment and Depression in Later Life
TABLE 2.
Multivariate Associations of Built Environment Attributes With Depression (Significant Effects Highlighted in Bold)
Variables Built environment Walkability
Street connectivity
Residential density
Land-use mix
Neighborhood composition SEIFA disadvantagea
Individual sociodemographics Age group
High school education Migrant Living alone Housing type
Individual psychosocial factors DSSI (social support)
APGAR (family and friends)
Sense of community
Individual health Charlson comorbidity
Smoking
Level
Model 2a, OR (95% CI)
Model 2b, OR (95% CI)
Model 2c, OR (95% CI)
Model 2d, OR (95% CI)
Model 2e, OR (95% CI)
1 1.53 (1.10–2.14) 1.53 (1.09–2.14)
1 0.83 (0.57–1.21) 0.93 (0.65–1.34) 0.79 (0.54–1.16) 1 1.12 (0.75–1.68) 1.31 (0.89–1.93) 1.20 (0.81–1.80) 1 1.54 (1.10–2.16) 1.52 (1.08–2.14)
Q1 (low) Q2 Q3 Q4 (high) Q1 (low) Q2 Q3 Q4 (high) Q1 (low) Q2 Q3 Q4 (high) T1 (low) T2 T3 (high)
1 1.11 (0.76–1.62) 1.23 (0.85–1.80) 1.14 (0.78–1.67)
T1 (low) T2 T3 (high)
1 0.91 (0.67–1.25) 0.97 (0.69–1.37)
1 0.91 (0.66–1.24) 0.99 (0.70–1.39)
1 0.90 (0.66–1.23) 0.98 (0.69–1.38)
1 0.96 (0.70–1.31) 1.02 (0.72–1.45)
1 0.93 (0.68–1.28) 1.04 (0.73–1.47)
69–74 75–79 80–84 85+ No Yes No Yes No Yes Own home Flat/unit Other
1 1.43 (1.05–1.93) 1.51 (1.03–2.22) 2.93 (1.58–5.44) 1 0.67 (0.51–0.89) 1 1.02 (0.78–1.34) 1 1.04 (0.74–1.46) 1 1.19 (0.82–1.73) 1.16 (0.74–1.81)
1 1.44 (1.06–1.95) 1.50 (1.02–2.21) 2.93 (1.58–5.43) 1 0.67 (0.51–0.89) 1 1.02 (0.78–1.34) 1 1.05 (0.75–1.47) 1 1.18 (0.81–1.71) 1.15 (0.73–1.80)
1 1.42 (1.05–1.93) 1.50 (1.02–2.21) 2.91 (1.57–5.42) 1 0.68 (0.51–0.90) 1 1.02 (0.78–1.34) 1 1.03 (0.73–1.45) 1 1.14 (0.78–1.67) 1.12 (0.71–1.76)
1 1.42 (1.05–1.92) 1.46 (0.99–2.14) 2.81 (1.51–5.23) 1 0.68 (0.51–0.90) 1 1.03 (0.78–1.35) 1 1.02 (0.73–1.44) 1 1.18 (0.81–1.71) 1.11 (0.71–1.75)
1 1.41 (1.04–1.91) 1.44 (0.98–2.12) 2.78 (1.49–5.19) 1 0.69 (0.52–0.91) 1 1.03 (0.79–1.35) 1 1.02 (0.73–1.44) 1 1.14 (0.78–1.67) 1.10 (0.69–1.73)
Low Medium High T1 (low) T2 T3 (high) T1 (low) T2 T3 (high)
1 0.30 (0.22–0.40) 0.09 (0.05–0.14) 1 0.52 (0.37–0.73) 0.62 (0.39–0.97) 1 0.79 (0.58–1.07) 0.79 (0.55–1.15)
1 0.30 (0.22–0.40) 0.09 (0.05–0.14) 1 0.52 (0.37–0.73) 0.62 (0.39–0.97) 1 0.78 (0.58–1.07) 0.80 (0.55–1.15)
1 0.30 (0.22–0.41) 0.09 (0.05–0.14) 1 0.52 (0.37–0.73) 0.61 (0.39–0.96) 1 0.79 (0.58–1.07) 0.79 (0.55–1.15)
1 0.30 (0.22–0.40) 0.09 (0.05–0.14) 1 0.52 (0.37–0.73) 0.61 (0.39–0.96) 1 0.77 (0.57–1.05) 0.77 (0.53–1.12)
1 0.30 (0.22–0.40) 0.09 (0.05–0.14) 1 0.52 (0.37–0.73) 0.61 (0.39–0.96) 1 0.77 (0.56–1.05) 0.78 (0.54–1.13)
0 1–2 3–4 5+ Never Past Current
1 1.47 (1.01–2.15) 2.35 (1.62–3.39) 3.17 (2.01–5.00) 1 1.50 (1.08–2.08) 2.32 (1.34–4.02)
1 1.48 (1.01–2.15) 2.38 (1.64–3.43) 3.13 (1.99–4.94) 1 1.50 (1.09–2.08) 2.33 (1.34–4.03)
1 1.48 (1.02–2.16) 2.34 (1.62–3.39) 3.14 (1.99–4.95) 1 1.51 (1.09–2.09) 2.39 (1.38–4.15)
1 1.49 (1.03–2.18) 2.40 (1.66–3.47) 3.20 (2.03–5.06) 1 1.51 (1.09–2.09) 2.34 (1.35–4.06)
1 1.50 (1.03–2.18) 2.41 (1.66–3.49) 3.21 (2.03–5.07) 1 1.52 (1.10–2.11) 2.38 (1.37–4.13)
1 0.86 (0.59–1.24) 0.97 (0.68–1.38) 0.84 (0.57–1.22) 1 1.03 (0.70–1.53) 1.24 (0.85–1.81) 1.21 (0.82–1.80)
Notes: Data are given as OR with 95% CIs. Q: Quartile; T: Tertile; SEIFA: Socioeconomic Index for Areas. a Higher values on the SEIFA disadvantage index imply less disadvantage.
466
Am J Geriatr Psychiatry 19:5, May 2011
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Am J Geriatr Psychiatry 19:5, May 2011 0 1–2 3–4 5+ Never Past Current
Low Medium High T1 (low) T2 T3 (high) T1 (low) T2 T3 (high)
69–74 75–79 80–84 85+ No Yes No Yes No Yes Own home Flat/unit Other
T1 (low) T2 T3 (high)
1 1.48 (1.02–2.15) 2.33 (1.61–3.38) 3.23 (2.05–5.10) 1 1.48 (1.07–2.05) 2.36 (1.36–4.09)
1 0.30 (0.22–0.40) 0.09 (0.05–0.14) 1 0.52 (0.37–0.73) 0.62 (0.39–0.97) 1 0.78 (0.57–1.06) 0.77 (0.53–1.11)
1 1.40 (1.04–1.90) 1.46 (0.99–2.15) 2.84 (1.53–5.28) 1 0.68 (0.51–0.89) 1 1.02 (0.77–1.33) 1 1.03 (0.73–1.44) 1 1.16 (0.80–1.69) 1.11 (0.71–1.75)
1 0.93 (0.68–1.27) 1.01 (0.71–1.43)
1 1.46 (1.11–1.90)
Model 3a, OR (95% CI)
1 1.49 (1.02–2.17) 2.35 (1.63–3.40) 3.15 (2.00–4.97) 1 1.50 (1.08–2.07) 2.35 (1.36–4.07)
1 0.30 (0.22–0.40) 0.09 (0.05–0.14) 1 0.52 (0.37–0.73) 0.61 (0.39–0.96) 1 0.79 (0.58–1.08) 0.79 (0.55–1.15)
1 1.44 (1.06–1.94) 1.50 (1.02–2.20) 2.93 (1.58–5.43) 1 0.67 (0.51–0.89) 1 1.01 (0.77–1.32) 1 1.04 (0.74–1.45) 1 1.18 (0.82–1.71) 1.15 (0.73–1.80)
1 0.92 (0.67–1.25) 0.98 (0.70–1.39)
1 1.28 (0.86–1.92)
Model 3b, OR (95% CI)
1 1.49 (1.02–2.17) 2.36 (1.63–3.41) 3.17 (2.01–5.00) 1 1.49 (1.07–2.06) 2.35 (1.36–4.06)
1 0.30 (0.22–0.40) 0.09 (0.05–0.14) 1 0.52 (0.36–0.73) 0.61 (0.39–0.96) 1 0.79 (0.58–1.07) 0.79 (0.55–1.14)
1 1.43 (1.06–1.93) 1.49 (1.02–2.20) 2.92 (1.57–5.41) 1 0.68 (0.51–0.90) 1 1.02 (0.77–1.33) 1 1.03 (0.74–1.45) 1 1.18 (0.81–1.70) 1.14 (0.73–1.79)
1 0.92 (0.67–1.25) 0.98 (0.70–1.39)
1 1.21 (0.92–1.60)
Model 3c, OR (95% CI)
Notes: Data are given as OR with 95% CI. Q: Quartile; T: Tertile; SEIFA: Socioeconomic Index for Areas. a Higher values on the SEIFA disadvantage index imply less disadvantage.
Smoking
Individual health Charlson comorbidity
Sense of community
APGAR (family and friends)
Individual psychosocial factors DSSI (social support)
Housing type
Living alone
Migrant
High school education
Individual sociodemographics Age group
Neighborhood composition SEIFA disadvantagea
Entertainment/recreation/culture
Health/well-being/community services
Offices/business
Other retail
No Yes No Yes No Yes No Yes No Yes
Level
1 1.49 (1.03–2.17) 2.38 (1.65–3.44) 3.17 (2.01–4.99) 1 1.51 (1.09–2.09) 2.34 (1.35–4.05)
1 0.29 (0.22–0.40) 0.09 (0.05–0.14) 1 0.52 (0.36–0.73) 0.61 (0.39–0.96) 1 0.79 (0.58–1.08) 0.79 (0.55–1.14)
1 1.45 (1.07–1.95) 1.50 (1.02–2.21) 2.94 (1.59–5.44) 1 0.67 (0.51–0.89) 1 1.03 (0.78–1.35) 1 1.04 (0.74–1.46) 1 1.17 (0.81–1.70) 1.12 (0.71–1.76)
1 0.94 (0.69–1.29) 1.01 (0.71–1.43)
1 1.22 (0.93–1.59)
Model 3d, OR (95% CI)
Multivariate Associations of Land-Use Availability Attributes With Depression (Significant Effects Highlighted in Bold)
Land-use availability Retail
Variables
TABLE 3.
1 1.48 (1.01–2.15) 2.36 (1.63–3.41) 3.14 (1.99–4.95) 1 1.50 (1.08–2.08) 2.35 (1.36–4.07)
1 0.30 (0.22–0.40) 0.09 (0.05–0.14) 1 0.52 (0.37–0.73) 0.61 (0.39–0.96) 1 0.79 (0.58–1.07) 0.79 (0.54–1.14)
1 1.44 (1.06–1.94) 1.51 (1.03–2.22) 2.94 (1.59–5.46) 1 0.67 (0.51–0.89) 1 1.02 (0.78–1.34) 1 1.04 (0.74–1.46) 1 1.18 (0.81–1.71) 1.14 (0.73–1.79)
1 0.91 (0.67–1.24) 0.98 (0.69–1.38)
1 1.08 (0.75–1.55)
Model 3e, OR (95% CI)
1 1.50 (1.03–2.18) 2.34 (1.62–3.39) 3.25 (2.06–5.13) 1 1.48 (1.07–2.05) 2.35 (1.36–4.08)
1 0.30 (0.22–0.40) 0.09 (0.05–0.14) 1 0.52 (0.37–0.73) 0.61 (0.39–0.96) 1 0.78 (0.57–1.07) 0.78 (0.54–1.12)
1 1.41 (1.04–1.91) 1.45 (0.99–2.14) 2.84 (1.53–5.28) 1 0.68 (0.51–0.89) 1 1.02 (0.77–1.34) 1 1.03 (0.73–1.44) 1 1.17 (0.80–1.69) 1.11 (0.71–1.74)
1 0.95 (0.69–1.30) 1.02 (0.72–1.45)
1 1.40 (1.04–1.90) 1 1.07 (0.69–1.66) 1 1.06 (0.77–1.45) 1 1.11 (0.84–1.47) 1 0.90 (0.61–1.32)
Model 3f, OR (95% CI)
Saarloos et al
467
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Built Environment and Depression in Later Life 0], “questionable” [1–4], “mild/moderate” [5–9], and “severe depressive symptoms” [10+]). No significant differences were found.
CONCLUSIONS Our study found that the odds of depression in older men were higher in areas with more landuse diversity, independent of neighborhood composition, diverse individual-level factors, and other BE attributes (street connectivity and residential density). Further analyses showed that this seemed to be driven by the presence of retail. Although these results are possibly counterintuitive, facilities and services—particularly retail—often attract more strangers to a neighborhood and generate more traffic.17 Earlier studies45,46 have shown that such settings may contribute to decreasing the social interaction between neighbors while increasing stress. There is also evidence that retail is negatively related to neighborhood satisfaction.47 Moreover, retail usually requires large parking lots that especially older adults may experience as unsafe because of either traffic or incivilities. Parking lots have been shown to decrease the sense of community in a neighborhood.48 Notably, our sample consisted of only men. Their feeling about specific BE features might differ from women, and their interaction with the environment. For instance, among people age 65 years and older living in major cities in Australia, 84% of men have access to a motor vehicle to drive compared with only 55% of women.49 Having easier access to other places, men may be more prone to focus on the disturbing aspects of local resources. In conjunction with the presence of retail, several covariates were associated with depression. The association between sense of community and depression was not independent of other factors; it was largely accounted for by aspects of social support. In fact, other studies have illustrated the role of social support as a coping mechanism to buffer harmful effects of other factors, such as being more homebound.50 The reported study has several strengths. First, the study used a large-size sample derived from a community-representative population of older men, which enabled us to investigate small effect size associations. Second, the links between BE attributes and depression were examined in a more inclusive man-
468
ner than in previous studies, because a large number of measured factors were available that have not been taken into account in concert before (e.g., social support and physical comorbidity). In this respect, it is important to realize that people who are very sick may not be able to walk or network effectively with neighbors, even when the environment is conducive. Third, the study used objective (as opposed to subjective) information about the BE, which rules out the likely association between mental illness and perception of environmental characteristics. There are several limitations to these findings. First, we used CCDs as surrogates of participants’ neighborhood. The disadvantages of this method are known.14 For example, it crudely assumes as a participant lives in the center of a CCD as the area captured will nearly always differ from the area that actually affects people. This is of particular concern when studying neighborhood effects on participation in walking or other physical activity around the home. In this study, however, the focus was on depression without making a priori assumptions about the distance over which the BE may have had an impact. Second, data on some potentially relevant confounders were unavailable at or around the time of survey. For instance, there is growing epidemiological and experimental evidence of an antidepressant effect of physical activity.51 Third, although in most cases covariates were turned into ordered categories to preserve information, the categorization itself may be a source of residual confounding.52 Fourth, the cross-sectional design of our study does not allow us to dismiss the possibility of reverse causality, although we found no significant differences between participants with varying degrees of depression. Fifth, the diagnosis of depression was not based on a structured clinical assessment according to acceptable criteria (Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition or International Classification of Diseases-10), although several studies have shown that the approach we used has good sensitivity, specificity, and positive predictive value. Sixth, we did not have access to data to ascertain the exact number of retail establishments that may be critical in this association with depression (i.e., how many retail establishments it would require in an area to cause harm?). Finally, it could be considered a limitation that the study involved men, making it impossible to assess to what
Am J Geriatr Psychiatry 19:5, May 2011
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Saarloos et al extent the associations found may be applicable to older women as well. This is an issue that deserves further research. In conclusion, this study suggests that the BE may have significant effects on depression in older men, independent of neighborhood composition factors and sociodemographic, psychosocial, and health factors at the individual level. Land-use mix and particularly retail availability are often considered as factors that may contribute to public health as it creates places for people to walk to and meet with others.53 Our findings, however, indicate that the relationship seems more complex when considering
mental health as well. Because the effects seem to be negative, a careful consideration of these attributes is required when designing or retrofitting residential environments, so as to optimally balance the impacts on physical and mental health across various population segments. The authors thank Professors Konrad Janrozik, Paul Norman, Graeme J. Hankey, and Leon Flicker for granting access to the HIMS data used in this article. This work was supported by a National Health and Medical Research Council (NHMRC) Capacity Building Grant, number 458668 (to BGC and OPA); an NHMRC Senior Research Fellowship, number 503712 (to BGC); three NHMRC Project Grants, numbers 279408, 379600, and 403963 (to OPA); and an MBF Foundation Project Grant, number DS 080608 (to OPA).
References 1. World Health Organization: Men, Ageing and Health: Achieving Health Across the Life Span. Geneva, World Health Organization, 2001 2. Pirkis J, Pfaff J, Williamson M, et al: The community prevalence of depression in older Australians. J Affect Disord 2009; 115:54–61 3. Satariano W: Epidemiology of Aging: An Ecological Approach. Sudbury, MA, Jones & Bartlett Publishers, 2006 4. Satariano WA: The disabilities of aging: looking to the physical environment. Am J Public Health 1997; 87:331–332 5. Blazer DG: Depression and social support in late life: a clear but not obvious relationship. Aging Mental Health 2005; 497–499 6. Evans G: The built environment and mental health. J Urban Health 2003; 80:536–555 7. Strawbridge WJ, Deleger S, Roberts RE, et al: Physical activity reduces the risk of subsequent depression for older adults. Am J Epidemiol 2002; 156:328–334 8. Gilman SE, Abraham HD: A longitudinal study of the order of onset of alcohol dependence and major depression. Drug Alcohol Depend 2001; 63:277–286 9. Husky MM, Mazure CM, Paliwal P, et al: Gender differences in the comorbidity of smoking behavior and major depression. Drug Alcohol Depend 2008; 93:176–179 10. Kim D: Blues from the neighborhood? Neighborhood characteristics and depression. Epidemiol Rev 2008; 30:101–117 11. Hybels CF, Blazer DG, Pieper CF, et al: Sociodemographic characteristics of the neighborhood and depressive symptoms in older adults: using multilevel modeling in geriatric psychiatry. Am J Geriatr Psychiatry 2006; 14:498–506 12. Aneshensel CS, Wight RG, Miller-Martinez D, et al: Urban neighborhoods and depressive symptoms among older adults. J Gerontol B Psychol Sci Soc Sci 2007; 62:S52–S59 13. Oswald F, Hieber A, Wahl H-W, et al: Ageing and person– environment fit in different urban neighbourhoods. Eur J Ageing 2005; 2:88–97 14. Diez Roux AV: Invited commentary: places, people, and health. Am J Epidemiol 2002; 155:516–519 15. Evans GW, Kantrowitz E, Eshelman P: Housing quality and psychological well-being among the elderly population. J Gerontol B Psychol Sci Soc Sci 2002; 57:P381–P383
Am J Geriatr Psychiatry 19:5, May 2011
16. van der Meer MJ: The sociospatial diversity in the leisure activities of older people in the Netherlands. J Aging Stud 2008; 22:1–12 17. Glass TA, Balfour JL: Neighborhoods, aging, and functional limitations, in Neighborhoods and Health. Edited by Kawachi I, Berkman LF. New York, Oxford University Press, 2003, pp 303–334 18. Fonda SJ, Wallace RB, Herzog AR: Changes in driving patterns and worsening depressive symptoms among older adults. J Gerontol B Psychol Sci Soc Sci 2001; 56B:S343–S351 19. Berry HL: ‘Crowded suburbs’ and ‘killer cities’: a brief review of the relationship between urban environments and mental health. N S W Public Health Bull 2007; 18:222–227 20. Berke EM, Gottlieb LM, Moudon AV, et al: Protective association between neighborhood walkability and depression in older men. J Am Geriatr Soc 2007; 55:526–533 21. Lee C, Moudon AV: The 3Ds + R: quantifying land use and urban form correlates of walking. Transp Res Part D 2006; 11:204–215 22. Pollack CE, Von dem Knesebeck O: Social capital and health among the aged: comparisons between the United States and Germany. Health Place 2004; 10:383–391 23. George LK, Blazer DG, Hughes DC, et al: Social support and the outcome of major depression. Br J Psychiatry 1989; 154:478–485 24. Norman PE, Flicker L, Almeida OP, et al: Cohort profile: the Health In Men Study (HIMS). Int J Epidemiol 2009; 38:48–52 25. Sheikh JI, Yesavage JA: Geriatric Depression Scale (GDS): recent evidence and development of a shorter version. Clin Gerontol 1986; 5:165–173 26. Almeida OP, Almeida SA: Short versions of the geriatric depression scale: a study of their validity for the diagnosis of a major depressive episode according to ICD-10 and DSM-IV. Int J Geriatr Psychiatry 1999; 14:858–865 27. Brownson RC, Hoehner CM, Day K, et al: Measuring the built environment for physical activity: state of the science. Am J Prev Med 2009; 36:S99–S123.e112 28. Frank LD, Schmid TL, Sallis JF, et al: Linking objectively measured physical activity with objectively measured urban form: findings from SMARTRAQ. Am J Prev Med 2005; 28:117–125 29. Leslie E, Coffee N, Frank L, et al: Walkability of local communities: using geographic information systems to objectively
469
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Built Environment and Depression in Later Life
30.
31.
32.
33.
34.
35. 36.
37.
38. 39.
40.
assess relevant environmental attributes. Health Place 2007; 13:111–122 Frank LD, Andresen MA, Schmid TL: Obesity relationships with community design, physical activity, and time spent in cars. Am J Prev Med 2004; 27:87–96 Ewing R, Schmid T, Killingsworth R, et al: Relationship between urban sprawl and physical activity, obesity, and morbidity. Am J Health Promot 2003; 18:47–57 ABS: Census of Population and Housing—Socio-economic Indexes for Areas. Canberra, Australia, Australian Bureau of Statistics, 2003 Lawton MP, Moss M, Moles E: The suprapersonal neighborhood context of older people: age heterogeneity and well-being. Environ Behav 1984; 16:89–109 Subramanian SV, Kubzansky L, Berkman L, et al: Neighborhood effects on the self-rated health of elders: uncovering the relative importance of structural and service-related neighborhood environments. J Gerontol B Psychol Sci Soc Sci 2006; 61: S153–S160 ABS: 2001 Census of Population and Housing. Canberra, Australia, Australian Bureau of Statistics, 2001 Vink D, Aartsen MJ, Comijs HC, et al: Onset of anxiety and depression in the aging population: comparison of risk factors in a 9-year prospective study. Am J Geriatr Psychiatry 2009; 17:642–652 Oxman TE, Berkman LF, Kasl S, et al: Social support and depressive symptoms in the elderly. Am J Epidemiol 1992; 135: 356–368 Smilkstein: The family APGAR: a proposal for a family function test and its use by physicians. J Fam Pract 1978; 6:1231–1239 Young AF, Russell A, Powers JR: The sense of belonging to a neighbourhood: can it be measured and is it related to health and well being in older women? Soc Sci Med 2004; 59:2627–2637 Charlson ME, Pompei P, Ales KL, et al: A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987; 40:373–383
470
41. Quan H, Sundararajan V, Halfon P, et al: Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data. Med Care 2005; 43:1130–1139 42. Almeida OP, Flicker L, Norman P, et al: Association of cardiovascular risk factors and disease with depression in later life. Am J Geriatr Psychiatry 2007; 15:506–513 43. StataCorp LP: Stata 10.1, TX, StataCorp LP, 2008 44. Mair C, Roux AVD, Galea S: Are neighbourhood characteristics associated with depressive symptoms? A review of evidence. J Epidemiol Community Health 2008; 62:940–946 45. Appleyard D: Livable Streets. Berkeley, CA, University of California Press, 1981 46. Halpern D: Mental Health and the Built Environment: More Than Bricks and Mortar? London, UK, Taylor & Francis, 1995 47. Ellis CD, Lee S-W, Kweon B-S: Retail land use, neighborhood satisfaction and the urban forest: an investigation into the moderating and mediating effects of trees and shrubs. Landsc Urban Plan 2006; 74:70–78 48. Wood L, Frank LD, Giles-Corti B: Sense of community and its relationship with walking and neighborhood design. Soc Sci Med 2010; 70:1381–1390 49. AIHW: Older Australia at a Glance, 4th ed. Canberra, Australian Institute of Health and Welfare, 2007. 50. Choi NG, McDougall GJ: Comparison of depressive symptoms between homebound older adults and ambulatory older adults. Aging Mental Health 2007; 11:310–322 51. Taylor AH, Cable NT, Faulkner G, et al: Physical activity and older adults: a review of health benefits and the effectiveness of interventions. J Sports Sci 2004; 22:703–725 52. Kaufman JS, Cooper RS, McGee DL: Socioeconomic status and health in blacks and whites: the problem of residual confounding and the resiliency of race. Epidemiology 1997; 8:621–628 53. Sugiyama T, Ward Thompson C: Older people’s health, outdoor activity and supportiveness of neighbourhood environments. Landsc Urban Plan 2007; 83:168–175
Am J Geriatr Psychiatry 19:5, May 2011
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Objective: This study examined the impact of built environment (BE) attributes on depression in older men to determine whether associations were independent of neighborhood composition factors and sociodemographic, psychosocial, and health factors at the individual level. Methods: The authors used geocoded data from the Health in Men Study collected in Western Australia in 2001 (N = 5,218). Depression was measured using the self-rated 15-item Geriatric Depression Scale. Geographic Information Systems were used to objectively measure BE attributes. Univariate logistic regressions were applied to select relevant covariates. Multivariate logistic regressions were conducted to examine BE attributes both separately and conjointly. Results: Higher degrees of land-use mix were associated with higher odds of depression independent of other factors, including street connectivity and residential density (odds ratio = 1.54, 95% confidence interval [CI] = 1.10–2.16, and odds ratio = 1.52, 95% CI = 1.08–2.14 for the second and third tertiles, respectively). Further examination showed that retail availability was associated with a 40% increase in the odds of depression (95% CI = 4%–90%) independent of other factors, including availability of other land uses. Conclusions: The BE is independently associated with depression through land-use mix, and specifically through retail availability. Although local retail facilitates walking, our findings suggest that it may increase the odds of depression in older men. This requires further exploration but suggests the need for careful planning of retail in residential environments, particularly near housing for older adults. (Am J Geriatr Psychiatry 2011; 19:461–470) Key Words: Ageing, elderly, men, depression, depressive disorder, mood disorder, mental health, risk factors, built environment, social context, epidemiology
D
epression is the most common and disabling functional disorder among ageing men.1 In Australia, clinically significant depression affects about ∼8.6% of older men.2 Many factors have
been associated with late-life depression.3 Besides individual characteristics, it has been suggested that social, behavioral, and environmental factors play a role in predisposing, precipitating,
Received November 23, 2009; revised April 19, 2010; accepted April 20, 2010. From the Centre for the Built Environment and Health, School of Population Health (DS, BG-C, NM); Western Australian Centre for Health and Ageing, Centre for Medical Research (HA, OPA); School of Psychiatry and Clinical Neurosciences (HA, OPA), The University of Western Australia, Australia; and Department of Psychiatry, Royal Perth Hospital, Australia (OPA). Send correspondence and reprint requests to Osvaldo P. Almeida, M.D., Ph.D., F.R.A.N.Z.C.P., F.F.P.O.A., WA Centre for Health and Ageing (M573), The University of Western Australia, 35 Stirling Highway, Crawley, WA 6009, Australia. e-mail: [email protected] c 2011 American Association for Geriatric Psychiatry DOI: 10.1097/JGP.0b013e3181e9b9bf
Am J Geriatr Psychiatry 19:5, May 2011
461
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Built Environment and Depression in Later Life and perpetuating depressive symptoms.3–6 Although much attention has been given to the influences of impaired social support3,5 and unhealthy behaviors,7–9 insight into the contribution of environmental factors remains limited. Hitherto, most studies have generally found that depression is less common in neighborhoods of higher socioeconomic status and more common in areas with greater social disorder.10 Other studies, however, have questioned the significance of these relationships.11,12 This inconsistency may well be due to a lack of attempts to examine the physical characteristics of the environment.10 It is argued that the built (or physical) environment becomes more important in later life.13–15 When people age, their life sphere often contracts to their residential environment due to changes such as physical decline, retirement, or decreased access to transport.16–18 This may also shrink their social networks.17 Hence, especially older people’s health and well-being may be affected by the physical and social settings of their neighborhoods. As such, some authors have suggested that the built environment (BE) may contribute to late-life depression.19 In a pioneering study, Berke et al.20 reported that depression was less prevalent among older men living in more walkable neighborhoods. The use of a composite measure such as “walkability,” however, fails to reveal the specific features of the BE that matter most for depression. To facilitate translation into practice,21 it will be necessary to investigate BE attributes that are specific and modifiable. Furthermore, it seems imperative to do so while accounting for social factors. Numerous studies have stressed the importance of social support and sense of community as buffers to stressors that lead to depression, both in the overall population and in older people.5,22,23 To our knowledge, it has not yet been determined to what extent BE attributes are significantly associated with depression above and beyond social factors. We conducted this study to examine the impact of BE attributes on depression in older men, taking an ecological approach that accounts for compositional neighborhood factors and sociodemographic, psychosocial, and health factors at the individual level. The study was designed to ascertain whether associations between BE attributes and depression in older men are independent of these other factors.
462
METHODS Sample and Study Design The study utilized data collected in the Health in Men Study (HIMS). Full details of the study design have been published elsewhere.24 Briefly, HIMS arose out of a population-based randomized trial that targeted men age 65–79 years living in Perth metropolitan area (Western Australia) in 1996–1999. All men were identified based on the electoral roll (enrolment to vote being compulsory in Australia). A followup survey was undertaken in 2001–2004. Our sample was taken from this follow-up survey and consisted of 5,218 older men whose residential addresses were geocoded. Data Preparation Geographic information systems were used to measure BE attributes for each Census Collection District (CCD) with one or more participants. The Western Australia Department for Planning and Infrastructure supplied land-use point data from the Western Australian Land Information Authority. Cadastral data were used to identify all land parcels in residential use. Parcels that overlapped CCD boundaries were intersected, and only the area within a CCD was counted. The Australian Bureau of Statistics provided data on the number of dwellings at the CCD level. A process was developed in which, for each cadastral parcel, the main land use was determined given the land-use classes defined by Department for Planning and Infrastructure and based on rating assessments of the Valuer Generals Office and Reserve Reports from Landgate. For the purpose of this study, where parcels had multiple land uses, we selected the main land use in the following order of preference: “retail” (= activities involving the sale of goods from shops located separate to and/or in a shopping centre), “other retail” (= activities that by virtue of the scale and special nature of goods differ from regular shops, e.g., car sales yard and carpet showroom), “offices and business,” “welfare, health and community services” (= activities providing the community with a specific service, such as hospitals, schools, and religious activities), and “entertainment, recreation, culture.”
Am J Geriatr Psychiatry 19:5, May 2011
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Saarloos et al Measures The prevalence of depression was assessed using the self-rated 15-item Geriatric Depression Scale.25 Participants with a total score of 7 or more were considered to have clinically significant depressive symptoms.26 Several modifiable BE attributes were assessed according to current knowledge on environmenthealth relationships.27–29 We measured street connectivity, residential density, and land-use mix in each person’s neighborhood. Street connectivity is an indicator of the interconnectedness of the street network in an area.28,30 We measured the number of intersections (nodes where three or more street segments meet) in a CCD divided by the area of that CCD (in km2 ). Residential density indicates the average density of residential developments in an area.30,31 This was measured as the total number of dwellings in a CCD divided by the total area of parcels in residential use within that CCD (in hectares). Land-use mix is an indicator of the diversity of land uses in an area.28 We assessed the five land-use classes mentioned and created an index with values ranging from zero (i.e., none of the land uses present) to one (i.e., an even mix of all five land uses). To evaluate neighborhood walkability, we standardized and summed the three measures for each area. Finally, we created measures indicating the availability of each land use in a CCD. Four categories of covariates were ascertained. First, factors of neighborhood composition were derived from census data.10 We obtained an index of relative socioeconomic disadvantage from the Socioeconomic Index for Areas 2001 dataset.32 This index takes into account area attributes such as low income, low educational attainment, unemployment, unskilled jobs, and variables that reflect disadvantage (e.g., indigenous and divorced). The higher a CCDs score on this index, the less disadvantaged it is compared with other areas. To investigate influences of age composition,33,34 we identified the percentages of children (0–14 years), younger adults (15– 24 years), and older adults (65 years and older) in each CCD from census data.35 Second, several individual sociodemographic characteristics were identified.10,36 We retrieved information from participants about their age, place of birth, and education level. We also asked them about their living arrange-
Am J Geriatr Psychiatry 19:5, May 2011
ments (With whom do you live now? [alone/with spouse only/with other relatives/with other nonrelatives/nursing home or hostel/other]), and their type of housing (In what kind of building do you live? [own home/flat, unit/elderly unit/nursing home, or hostel/other]). Third, several psychosocial factors were measured.5,37 We used the Duke Social Support Index (DSSI) to measure people’s satisfaction with their network of relationships. The DSSI generates scores ranging from 6 to 18, with higher scores indicating greater satisfaction. The Family and Friends Adaptation, Partnership, Growth, Affection, Resolve (APGAR) Scale38 was used to assess how participants judge the functioning of their social network, where higher scores suggest better functioning. Also, a 13-item measure of sense of community was included.39 The fourth category of covariates involved individual health factors. Physical comorbidity was measured with the weighted Charlson Comorbidity Index,40 which takes into account both number and seriousness of 17 common medical conditions that predict 1-year mortality. Comorbidity data were retrieved from the Western Australia Data Linkage System using established coding algorithms.41 Also, participants were asked about their smoking behavior (Have you ever smoked regularly? [yes/no]; if yes, how often do you smoke now? [every day/not every day/not at all]) at the time of assessment.42 Statistical Analysis BE attributes were transformed into categorical variables guided by the distribution of observed values. The five land-use availability attributes were measured as binary variables, indicating whether or not any part of a participant’s CCD was occupied by a specific land use. Land-use mix was converted into tertiles, whereas walkability, street connectivity, and residential density were split into quartiles. All covariates measuring neighborhood composition were turned into tertiles. Individual sociodemographic covariates were categorized into groups with sufficient observations. In case of DSSI, we allocated participants to three groups: 6–14 (lowest), 15–17 (middle), and 18 (highest). The APGAR Scale and sense of community measure were transformed into tertiles. Comorbidity and smoking behavior were operationalized into legitimate categories.
463
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Built Environment and Depression in Later Life We used Stata 10.1 software to analyze the data.43 Alpha was set at 5%, and all tests performed were two-tailed. The conceptual framework underpinning the analyses was based on the assumption that characteristics of the BE that facilitate social interactions decrease the risk of depression and that demographic, health, and psychosocial factors may confound this association. Therefore, we completed the analyses in three steps. First, univariate logistic regressions were conducted to assess the crude associations of walkability and its three components (street connectivity, residential density, and land-use mix) with depression and to select covariates significantly associated with depression. Second, multivariate logistic regressions were applied to further examine the BE attributes both separately and conjointly, while adjusting for selected covariates. Third, we unpacked the land-use mix variable to investigate the role of specific land uses by means of multivariate logistic regressions.
RESULTS A total of 295 men (5.7%) reported clinically significant depressive symptoms. Table 1 shows the univariate associations of all BE attributes and covariates with depression. Among the BE attributes, residential density was associated with depression, because those in the highest quartile were more likely to be depressed (odds ratio [OR] = 1.44; [95% confidence interval {CI} = 1.03–2.02]). The same trend was found for land-use mix, with significantly higher odds found for the highest tertile (OR = 1.37 [95% CI = 1.02–1.84]). Among factors of neighborhood composition, the Socioeconomic Index for Areas disadvantage index showed that the odds of depression were significantly lower in the least disadvantaged areas (OR = 0.69 [95% CI = 0.51–0.92]). Regarding individual sociodemographics, higher odds of depression were found among older men, those born overseas, those living alone, and those not living in their own home. Conversely, the odds of depression were lower in men who had an education level of at least high school. All included psychosocial factors revealed inverse relationships with depression. The odds were significantly lower in the highest tertiles of DSSI (OR = 0.05 [95% CI = 0.03–0.07]), APGAR (OR = 0.19 [95% CI = 0.13–0.28]), and sense of community
464
(OR = 0.34 [95% CI = 0.25–0.47]). With respect to health factors, we found higher odds of depression related to higher comorbidity and smoking in past and present. Next, we investigated the association between depression and the three components that make up the walkability index, controlling for covariates that showed a significant univariate association with depression (Table 2). Only land-use mix was significantly associated with depression independent of neighborhood composition and individual level factors. After additionally controlling for the effects of street connectivity and residential density (Model 2e), older men living in areas with the two highest tertiles of land-use mix were on average 53% more likely to be depressed than those in the lowest tertile of land-use diversity. When examined, 89% of areas in the lowest tertile contained none of the land uses, whereas 97% of areas in the middle tertile had one or two land uses present, and all areas in the highest tertile included two or more land uses. Compared with the covariates, the effect of land-use mix on depression was similar in size to the effects of educational attainment and APGAR. The strongest was the effect of DSSI with the odds of depression being 11 times lower in those being most satisfied (OR = 0.09 [95% CI = 0.05–0.14]). At an intermediate level, age, comorbidity, and smoking behavior were related to increased odds of depression. Finally, the land-use mix indicator was dissected into five land-use availability variables to examine the effects of specific land useson in older men with retail in their CCD were higher (OR = 1.46 [95% CI = 1.11–1.90]) than in areas without retail. There was no interaction between land-use mix and retail availability. When additionally controlling for other landuse availability variables (Model 3f), the effect of retail availability remained significant (OR = 1.40 [95% CI = 1.04–1.90]). We also examined the availability of parks and gardens, a subset of “entertainment/ recreation/culture,” but its effect on depression was insignificant. Reverse causality may exist in the sense that people more prone to depression may choose to live in neighborhoods with mixed land use or with retail. To assess whether this “social drift”10,19,44 may have occurred in our sample, we compared the results for participants with varying degrees of depression (“no depression” [Geriatric Depression Scale score =
Am J Geriatr Psychiatry 19:5, May 2011
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Saarloos et al
TABLE 1.
Univariate Associations With Depression (Significant Effects Highlighted in Bold)
Variables Built environment Walkability
Street connectivity
Residential density
Land-use mix
Neighborhood composition SEIFA disadvantagea
Percentage of children (0–14 years)
Percentage of younger adults (15–24 years)
Percentage of older adults (65+ years)
Individual sociodemographics Age group (years)
High school education Migrant Living alone Housing type
Individual psychosocial factors DSSI (social support)
APGAR (family and friends)
Sense of community
Individual health Charlson comorbidity
Smoking
Total
Depression (n = 295)
No Depression (n = 4,923)
Odds Ratio (95% CI)
Q1 (low) Q2 Q3 Q4 (high) Q1 (low) Q2 Q3 Q4 (high) Q1 (low) Q2 Q3 Q4 (high) T1 (low) T2 T3 (high)
1,317 1,302 1,308 1,291 1,340 1,282 1,352 1,244 1,314 1,314 1,302 1,288 1,749 1,744 1,725
63 (21.4) 78 (26.4) 75 (25.4) 79 (26.8) 81 (27.5) 68 (23.1) 78 (26.4) 68 (23.1) 62 (21.0) 65 (22.0) 82 (27.8) 86 (29.2) 82 (27.8) 104 (35.3) 109 (36.9)
1,254 (25.5) 1,224 (24.9) 1,233 (25.0) 1,212 (24.6) 1,259 (25.6) 1,214 (24.7) 1,274 (25.9) 1,176 (23.9) 1,252 (25.4) 1,249 (25.4) 1,220 (24.8) 1,202 (24.4) 1,667 (33.9) 1,640 (33.3) 1,616 (32.8)
1 1.27 (0.90–1.78) 1.21 (0.86–1.71) 1.30 (0.92–1.82) 1 0.87 (0.62–1.21) 0.95 (0.69–1.31) 0.90 (0.64–1.25) 1 1.05 (0.74–1.50) 1.36 (0.97–1.91) 1.44 (1.03–2.02) 1 1.29 (0.96–1.74) 1.37 (1.02–1.84)
T1 (low) T2 T3 (high) T1 (low) T2 T3 (high) T1 (low) T2 T3 (high) T1 (low) T2 T3 (high)
1,713 1,707 1,748 1,731 1,742 1,734 1,754 1,713 1,740 1,746 1,727 1,734
116 (39.3) 94 (31.9) 83 (28.1) 103 (34.9) 95 (32.2) 97 (32.9) 110 (37.3) 94 (31.9) 91 (30.8) 100 (33.9) 84 (28.5) 111 (37.6)
1,597 (32.4) 1,613 (32.8) 1,665 (33.8) 1,628 (33.1) 1,647 (33.5) 1,637 (33.3) 1,644 (33.4) 1,619 (32.9) 1,649 (33.5) 1,646 (33.4) 1,643 (33.4) 1,623 (33.0)
1 0.80 (0.61–1.06) 0.69 (0.51–0.92) 1 0.91 (0.68–1.21) 0.94 (0.70–1.25) 1 0.87 (0.65–1.15) 0.82 (0.62–1.10) 1 0.84 (0.62–1.13) 1.13 (0.85–1.49)
69–74 75–79 80–84 85+ No Yes No Yes No Yes Own home Flat/unit Other
2,465 1,752 821 160 2,724 2,252 2,972 2,006 4,355 843 4,099 709 406
106 (35.9) 114 (38.6) 57 (19.3) 17 (5.8) 185 (62.7) 98 (33.2) 148 (50.2) 135 (45.8) 232 (78.6) 62 (21.0) 210 (71.2) 52 (17.6) 32 (10.8)
2,359 (47.9) 1,638 (33.3) 764 (15.5) 143 (2.9) 2,539 (51.6) 2,154 (43.8) 2,824 (57.4) 1,871 (38.0) 4,123 (83.7) 781 (15.9) 3,889 (79.0) 657 (13.3) 374 (7.6)
1 1.55 (1.18–2.03) 1.66 (1.19–2.31) 2.65 (1.54–4.54) 1 0.62 (0.49–0.80) 1 1.38 (1.08–1.75) 1 1.41 (1.06–1.89) 1 1.47 (1.07–2.01) 1.58 (1.08–2.33)
Low Medium High T1 (low) T2 T3 (high) T1 (low) T2 T3 (high)
875 2,087 2,226 1,823 1,847 1,482 1,728 1,771 1,672
166 (56.3) 100 (33.9) 25 (8.5) 196 (66.4) 57 (19.3) 33 (11.2) 150 (50.8) 86 (29.2) 53 (18.0)
709 (14.4) 1,987 (40.4) 2,201 (44.7) 1,627 (33.0) 1,790 (36.4) 1,449 (29.4) 1,578 (32.1) 1,685 (34.2) 1,619 (32.9)
1 0.21 (0.17–0.28) 0.05 (0.03–0.07) 1 0.26 (0.20–0.36) 0.19 (0.13–0.28) 1 0.54 (0.41–0.71) 0.34 (0.25–0.47)
0 1–2 3–4 5+ Never Past Current
3,807 686 484 241 1,703 3,228 264
161 (54.6) 48 (16.3) 52 (17.6) 34 (11.5) 59 (20.0) 209 (70.8) 26 (8.8)
3,646 (74.1) 638 (13.0) 432 (8.8) 207 (4.2) 1,644 (33.4) 3,019 (61.3) 238 (4.8)
1 1.70 (1.22–2.38) 2.73 (1.96–3.78) 3.72 (2.50–5.52) 1 1.93 (1.44–2.59) 3.04 (1.88–4.92)
Level
Notes: Data are given as N (%) or OR with 95% CIs. Q: Quartile; T: Tertile. SEIFA: Socioeconomic Index for Areas. a Higher values on the SEIFA disadvantage index imply less disadvantage.
Am J Geriatr Psychiatry 19:5, May 2011
465
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Built Environment and Depression in Later Life
TABLE 2.
Multivariate Associations of Built Environment Attributes With Depression (Significant Effects Highlighted in Bold)
Variables Built environment Walkability
Street connectivity
Residential density
Land-use mix
Neighborhood composition SEIFA disadvantagea
Individual sociodemographics Age group
High school education Migrant Living alone Housing type
Individual psychosocial factors DSSI (social support)
APGAR (family and friends)
Sense of community
Individual health Charlson comorbidity
Smoking
Level
Model 2a, OR (95% CI)
Model 2b, OR (95% CI)
Model 2c, OR (95% CI)
Model 2d, OR (95% CI)
Model 2e, OR (95% CI)
1 1.53 (1.10–2.14) 1.53 (1.09–2.14)
1 0.83 (0.57–1.21) 0.93 (0.65–1.34) 0.79 (0.54–1.16) 1 1.12 (0.75–1.68) 1.31 (0.89–1.93) 1.20 (0.81–1.80) 1 1.54 (1.10–2.16) 1.52 (1.08–2.14)
Q1 (low) Q2 Q3 Q4 (high) Q1 (low) Q2 Q3 Q4 (high) Q1 (low) Q2 Q3 Q4 (high) T1 (low) T2 T3 (high)
1 1.11 (0.76–1.62) 1.23 (0.85–1.80) 1.14 (0.78–1.67)
T1 (low) T2 T3 (high)
1 0.91 (0.67–1.25) 0.97 (0.69–1.37)
1 0.91 (0.66–1.24) 0.99 (0.70–1.39)
1 0.90 (0.66–1.23) 0.98 (0.69–1.38)
1 0.96 (0.70–1.31) 1.02 (0.72–1.45)
1 0.93 (0.68–1.28) 1.04 (0.73–1.47)
69–74 75–79 80–84 85+ No Yes No Yes No Yes Own home Flat/unit Other
1 1.43 (1.05–1.93) 1.51 (1.03–2.22) 2.93 (1.58–5.44) 1 0.67 (0.51–0.89) 1 1.02 (0.78–1.34) 1 1.04 (0.74–1.46) 1 1.19 (0.82–1.73) 1.16 (0.74–1.81)
1 1.44 (1.06–1.95) 1.50 (1.02–2.21) 2.93 (1.58–5.43) 1 0.67 (0.51–0.89) 1 1.02 (0.78–1.34) 1 1.05 (0.75–1.47) 1 1.18 (0.81–1.71) 1.15 (0.73–1.80)
1 1.42 (1.05–1.93) 1.50 (1.02–2.21) 2.91 (1.57–5.42) 1 0.68 (0.51–0.90) 1 1.02 (0.78–1.34) 1 1.03 (0.73–1.45) 1 1.14 (0.78–1.67) 1.12 (0.71–1.76)
1 1.42 (1.05–1.92) 1.46 (0.99–2.14) 2.81 (1.51–5.23) 1 0.68 (0.51–0.90) 1 1.03 (0.78–1.35) 1 1.02 (0.73–1.44) 1 1.18 (0.81–1.71) 1.11 (0.71–1.75)
1 1.41 (1.04–1.91) 1.44 (0.98–2.12) 2.78 (1.49–5.19) 1 0.69 (0.52–0.91) 1 1.03 (0.79–1.35) 1 1.02 (0.73–1.44) 1 1.14 (0.78–1.67) 1.10 (0.69–1.73)
Low Medium High T1 (low) T2 T3 (high) T1 (low) T2 T3 (high)
1 0.30 (0.22–0.40) 0.09 (0.05–0.14) 1 0.52 (0.37–0.73) 0.62 (0.39–0.97) 1 0.79 (0.58–1.07) 0.79 (0.55–1.15)
1 0.30 (0.22–0.40) 0.09 (0.05–0.14) 1 0.52 (0.37–0.73) 0.62 (0.39–0.97) 1 0.78 (0.58–1.07) 0.80 (0.55–1.15)
1 0.30 (0.22–0.41) 0.09 (0.05–0.14) 1 0.52 (0.37–0.73) 0.61 (0.39–0.96) 1 0.79 (0.58–1.07) 0.79 (0.55–1.15)
1 0.30 (0.22–0.40) 0.09 (0.05–0.14) 1 0.52 (0.37–0.73) 0.61 (0.39–0.96) 1 0.77 (0.57–1.05) 0.77 (0.53–1.12)
1 0.30 (0.22–0.40) 0.09 (0.05–0.14) 1 0.52 (0.37–0.73) 0.61 (0.39–0.96) 1 0.77 (0.56–1.05) 0.78 (0.54–1.13)
0 1–2 3–4 5+ Never Past Current
1 1.47 (1.01–2.15) 2.35 (1.62–3.39) 3.17 (2.01–5.00) 1 1.50 (1.08–2.08) 2.32 (1.34–4.02)
1 1.48 (1.01–2.15) 2.38 (1.64–3.43) 3.13 (1.99–4.94) 1 1.50 (1.09–2.08) 2.33 (1.34–4.03)
1 1.48 (1.02–2.16) 2.34 (1.62–3.39) 3.14 (1.99–4.95) 1 1.51 (1.09–2.09) 2.39 (1.38–4.15)
1 1.49 (1.03–2.18) 2.40 (1.66–3.47) 3.20 (2.03–5.06) 1 1.51 (1.09–2.09) 2.34 (1.35–4.06)
1 1.50 (1.03–2.18) 2.41 (1.66–3.49) 3.21 (2.03–5.07) 1 1.52 (1.10–2.11) 2.38 (1.37–4.13)
1 0.86 (0.59–1.24) 0.97 (0.68–1.38) 0.84 (0.57–1.22) 1 1.03 (0.70–1.53) 1.24 (0.85–1.81) 1.21 (0.82–1.80)
Notes: Data are given as OR with 95% CIs. Q: Quartile; T: Tertile; SEIFA: Socioeconomic Index for Areas. a Higher values on the SEIFA disadvantage index imply less disadvantage.
466
Am J Geriatr Psychiatry 19:5, May 2011
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Am J Geriatr Psychiatry 19:5, May 2011 0 1–2 3–4 5+ Never Past Current
Low Medium High T1 (low) T2 T3 (high) T1 (low) T2 T3 (high)
69–74 75–79 80–84 85+ No Yes No Yes No Yes Own home Flat/unit Other
T1 (low) T2 T3 (high)
1 1.48 (1.02–2.15) 2.33 (1.61–3.38) 3.23 (2.05–5.10) 1 1.48 (1.07–2.05) 2.36 (1.36–4.09)
1 0.30 (0.22–0.40) 0.09 (0.05–0.14) 1 0.52 (0.37–0.73) 0.62 (0.39–0.97) 1 0.78 (0.57–1.06) 0.77 (0.53–1.11)
1 1.40 (1.04–1.90) 1.46 (0.99–2.15) 2.84 (1.53–5.28) 1 0.68 (0.51–0.89) 1 1.02 (0.77–1.33) 1 1.03 (0.73–1.44) 1 1.16 (0.80–1.69) 1.11 (0.71–1.75)
1 0.93 (0.68–1.27) 1.01 (0.71–1.43)
1 1.46 (1.11–1.90)
Model 3a, OR (95% CI)
1 1.49 (1.02–2.17) 2.35 (1.63–3.40) 3.15 (2.00–4.97) 1 1.50 (1.08–2.07) 2.35 (1.36–4.07)
1 0.30 (0.22–0.40) 0.09 (0.05–0.14) 1 0.52 (0.37–0.73) 0.61 (0.39–0.96) 1 0.79 (0.58–1.08) 0.79 (0.55–1.15)
1 1.44 (1.06–1.94) 1.50 (1.02–2.20) 2.93 (1.58–5.43) 1 0.67 (0.51–0.89) 1 1.01 (0.77–1.32) 1 1.04 (0.74–1.45) 1 1.18 (0.82–1.71) 1.15 (0.73–1.80)
1 0.92 (0.67–1.25) 0.98 (0.70–1.39)
1 1.28 (0.86–1.92)
Model 3b, OR (95% CI)
1 1.49 (1.02–2.17) 2.36 (1.63–3.41) 3.17 (2.01–5.00) 1 1.49 (1.07–2.06) 2.35 (1.36–4.06)
1 0.30 (0.22–0.40) 0.09 (0.05–0.14) 1 0.52 (0.36–0.73) 0.61 (0.39–0.96) 1 0.79 (0.58–1.07) 0.79 (0.55–1.14)
1 1.43 (1.06–1.93) 1.49 (1.02–2.20) 2.92 (1.57–5.41) 1 0.68 (0.51–0.90) 1 1.02 (0.77–1.33) 1 1.03 (0.74–1.45) 1 1.18 (0.81–1.70) 1.14 (0.73–1.79)
1 0.92 (0.67–1.25) 0.98 (0.70–1.39)
1 1.21 (0.92–1.60)
Model 3c, OR (95% CI)
Notes: Data are given as OR with 95% CI. Q: Quartile; T: Tertile; SEIFA: Socioeconomic Index for Areas. a Higher values on the SEIFA disadvantage index imply less disadvantage.
Smoking
Individual health Charlson comorbidity
Sense of community
APGAR (family and friends)
Individual psychosocial factors DSSI (social support)
Housing type
Living alone
Migrant
High school education
Individual sociodemographics Age group
Neighborhood composition SEIFA disadvantagea
Entertainment/recreation/culture
Health/well-being/community services
Offices/business
Other retail
No Yes No Yes No Yes No Yes No Yes
Level
1 1.49 (1.03–2.17) 2.38 (1.65–3.44) 3.17 (2.01–4.99) 1 1.51 (1.09–2.09) 2.34 (1.35–4.05)
1 0.29 (0.22–0.40) 0.09 (0.05–0.14) 1 0.52 (0.36–0.73) 0.61 (0.39–0.96) 1 0.79 (0.58–1.08) 0.79 (0.55–1.14)
1 1.45 (1.07–1.95) 1.50 (1.02–2.21) 2.94 (1.59–5.44) 1 0.67 (0.51–0.89) 1 1.03 (0.78–1.35) 1 1.04 (0.74–1.46) 1 1.17 (0.81–1.70) 1.12 (0.71–1.76)
1 0.94 (0.69–1.29) 1.01 (0.71–1.43)
1 1.22 (0.93–1.59)
Model 3d, OR (95% CI)
Multivariate Associations of Land-Use Availability Attributes With Depression (Significant Effects Highlighted in Bold)
Land-use availability Retail
Variables
TABLE 3.
1 1.48 (1.01–2.15) 2.36 (1.63–3.41) 3.14 (1.99–4.95) 1 1.50 (1.08–2.08) 2.35 (1.36–4.07)
1 0.30 (0.22–0.40) 0.09 (0.05–0.14) 1 0.52 (0.37–0.73) 0.61 (0.39–0.96) 1 0.79 (0.58–1.07) 0.79 (0.54–1.14)
1 1.44 (1.06–1.94) 1.51 (1.03–2.22) 2.94 (1.59–5.46) 1 0.67 (0.51–0.89) 1 1.02 (0.78–1.34) 1 1.04 (0.74–1.46) 1 1.18 (0.81–1.71) 1.14 (0.73–1.79)
1 0.91 (0.67–1.24) 0.98 (0.69–1.38)
1 1.08 (0.75–1.55)
Model 3e, OR (95% CI)
1 1.50 (1.03–2.18) 2.34 (1.62–3.39) 3.25 (2.06–5.13) 1 1.48 (1.07–2.05) 2.35 (1.36–4.08)
1 0.30 (0.22–0.40) 0.09 (0.05–0.14) 1 0.52 (0.37–0.73) 0.61 (0.39–0.96) 1 0.78 (0.57–1.07) 0.78 (0.54–1.12)
1 1.41 (1.04–1.91) 1.45 (0.99–2.14) 2.84 (1.53–5.28) 1 0.68 (0.51–0.89) 1 1.02 (0.77–1.34) 1 1.03 (0.73–1.44) 1 1.17 (0.80–1.69) 1.11 (0.71–1.74)
1 0.95 (0.69–1.30) 1.02 (0.72–1.45)
1 1.40 (1.04–1.90) 1 1.07 (0.69–1.66) 1 1.06 (0.77–1.45) 1 1.11 (0.84–1.47) 1 0.90 (0.61–1.32)
Model 3f, OR (95% CI)
Saarloos et al
467
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Built Environment and Depression in Later Life 0], “questionable” [1–4], “mild/moderate” [5–9], and “severe depressive symptoms” [10+]). No significant differences were found.
CONCLUSIONS Our study found that the odds of depression in older men were higher in areas with more landuse diversity, independent of neighborhood composition, diverse individual-level factors, and other BE attributes (street connectivity and residential density). Further analyses showed that this seemed to be driven by the presence of retail. Although these results are possibly counterintuitive, facilities and services—particularly retail—often attract more strangers to a neighborhood and generate more traffic.17 Earlier studies45,46 have shown that such settings may contribute to decreasing the social interaction between neighbors while increasing stress. There is also evidence that retail is negatively related to neighborhood satisfaction.47 Moreover, retail usually requires large parking lots that especially older adults may experience as unsafe because of either traffic or incivilities. Parking lots have been shown to decrease the sense of community in a neighborhood.48 Notably, our sample consisted of only men. Their feeling about specific BE features might differ from women, and their interaction with the environment. For instance, among people age 65 years and older living in major cities in Australia, 84% of men have access to a motor vehicle to drive compared with only 55% of women.49 Having easier access to other places, men may be more prone to focus on the disturbing aspects of local resources. In conjunction with the presence of retail, several covariates were associated with depression. The association between sense of community and depression was not independent of other factors; it was largely accounted for by aspects of social support. In fact, other studies have illustrated the role of social support as a coping mechanism to buffer harmful effects of other factors, such as being more homebound.50 The reported study has several strengths. First, the study used a large-size sample derived from a community-representative population of older men, which enabled us to investigate small effect size associations. Second, the links between BE attributes and depression were examined in a more inclusive man-
468
ner than in previous studies, because a large number of measured factors were available that have not been taken into account in concert before (e.g., social support and physical comorbidity). In this respect, it is important to realize that people who are very sick may not be able to walk or network effectively with neighbors, even when the environment is conducive. Third, the study used objective (as opposed to subjective) information about the BE, which rules out the likely association between mental illness and perception of environmental characteristics. There are several limitations to these findings. First, we used CCDs as surrogates of participants’ neighborhood. The disadvantages of this method are known.14 For example, it crudely assumes as a participant lives in the center of a CCD as the area captured will nearly always differ from the area that actually affects people. This is of particular concern when studying neighborhood effects on participation in walking or other physical activity around the home. In this study, however, the focus was on depression without making a priori assumptions about the distance over which the BE may have had an impact. Second, data on some potentially relevant confounders were unavailable at or around the time of survey. For instance, there is growing epidemiological and experimental evidence of an antidepressant effect of physical activity.51 Third, although in most cases covariates were turned into ordered categories to preserve information, the categorization itself may be a source of residual confounding.52 Fourth, the cross-sectional design of our study does not allow us to dismiss the possibility of reverse causality, although we found no significant differences between participants with varying degrees of depression. Fifth, the diagnosis of depression was not based on a structured clinical assessment according to acceptable criteria (Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition or International Classification of Diseases-10), although several studies have shown that the approach we used has good sensitivity, specificity, and positive predictive value. Sixth, we did not have access to data to ascertain the exact number of retail establishments that may be critical in this association with depression (i.e., how many retail establishments it would require in an area to cause harm?). Finally, it could be considered a limitation that the study involved men, making it impossible to assess to what
Am J Geriatr Psychiatry 19:5, May 2011
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Saarloos et al extent the associations found may be applicable to older women as well. This is an issue that deserves further research. In conclusion, this study suggests that the BE may have significant effects on depression in older men, independent of neighborhood composition factors and sociodemographic, psychosocial, and health factors at the individual level. Land-use mix and particularly retail availability are often considered as factors that may contribute to public health as it creates places for people to walk to and meet with others.53 Our findings, however, indicate that the relationship seems more complex when considering
mental health as well. Because the effects seem to be negative, a careful consideration of these attributes is required when designing or retrofitting residential environments, so as to optimally balance the impacts on physical and mental health across various population segments. The authors thank Professors Konrad Janrozik, Paul Norman, Graeme J. Hankey, and Leon Flicker for granting access to the HIMS data used in this article. This work was supported by a National Health and Medical Research Council (NHMRC) Capacity Building Grant, number 458668 (to BGC and OPA); an NHMRC Senior Research Fellowship, number 503712 (to BGC); three NHMRC Project Grants, numbers 279408, 379600, and 403963 (to OPA); and an MBF Foundation Project Grant, number DS 080608 (to OPA).
References 1. World Health Organization: Men, Ageing and Health: Achieving Health Across the Life Span. Geneva, World Health Organization, 2001 2. Pirkis J, Pfaff J, Williamson M, et al: The community prevalence of depression in older Australians. J Affect Disord 2009; 115:54–61 3. Satariano W: Epidemiology of Aging: An Ecological Approach. Sudbury, MA, Jones & Bartlett Publishers, 2006 4. Satariano WA: The disabilities of aging: looking to the physical environment. Am J Public Health 1997; 87:331–332 5. Blazer DG: Depression and social support in late life: a clear but not obvious relationship. Aging Mental Health 2005; 497–499 6. Evans G: The built environment and mental health. J Urban Health 2003; 80:536–555 7. Strawbridge WJ, Deleger S, Roberts RE, et al: Physical activity reduces the risk of subsequent depression for older adults. Am J Epidemiol 2002; 156:328–334 8. Gilman SE, Abraham HD: A longitudinal study of the order of onset of alcohol dependence and major depression. Drug Alcohol Depend 2001; 63:277–286 9. Husky MM, Mazure CM, Paliwal P, et al: Gender differences in the comorbidity of smoking behavior and major depression. Drug Alcohol Depend 2008; 93:176–179 10. Kim D: Blues from the neighborhood? Neighborhood characteristics and depression. Epidemiol Rev 2008; 30:101–117 11. Hybels CF, Blazer DG, Pieper CF, et al: Sociodemographic characteristics of the neighborhood and depressive symptoms in older adults: using multilevel modeling in geriatric psychiatry. Am J Geriatr Psychiatry 2006; 14:498–506 12. Aneshensel CS, Wight RG, Miller-Martinez D, et al: Urban neighborhoods and depressive symptoms among older adults. J Gerontol B Psychol Sci Soc Sci 2007; 62:S52–S59 13. Oswald F, Hieber A, Wahl H-W, et al: Ageing and person– environment fit in different urban neighbourhoods. Eur J Ageing 2005; 2:88–97 14. Diez Roux AV: Invited commentary: places, people, and health. Am J Epidemiol 2002; 155:516–519 15. Evans GW, Kantrowitz E, Eshelman P: Housing quality and psychological well-being among the elderly population. J Gerontol B Psychol Sci Soc Sci 2002; 57:P381–P383
Am J Geriatr Psychiatry 19:5, May 2011
16. van der Meer MJ: The sociospatial diversity in the leisure activities of older people in the Netherlands. J Aging Stud 2008; 22:1–12 17. Glass TA, Balfour JL: Neighborhoods, aging, and functional limitations, in Neighborhoods and Health. Edited by Kawachi I, Berkman LF. New York, Oxford University Press, 2003, pp 303–334 18. Fonda SJ, Wallace RB, Herzog AR: Changes in driving patterns and worsening depressive symptoms among older adults. J Gerontol B Psychol Sci Soc Sci 2001; 56B:S343–S351 19. Berry HL: ‘Crowded suburbs’ and ‘killer cities’: a brief review of the relationship between urban environments and mental health. N S W Public Health Bull 2007; 18:222–227 20. Berke EM, Gottlieb LM, Moudon AV, et al: Protective association between neighborhood walkability and depression in older men. J Am Geriatr Soc 2007; 55:526–533 21. Lee C, Moudon AV: The 3Ds + R: quantifying land use and urban form correlates of walking. Transp Res Part D 2006; 11:204–215 22. Pollack CE, Von dem Knesebeck O: Social capital and health among the aged: comparisons between the United States and Germany. Health Place 2004; 10:383–391 23. George LK, Blazer DG, Hughes DC, et al: Social support and the outcome of major depression. Br J Psychiatry 1989; 154:478–485 24. Norman PE, Flicker L, Almeida OP, et al: Cohort profile: the Health In Men Study (HIMS). Int J Epidemiol 2009; 38:48–52 25. Sheikh JI, Yesavage JA: Geriatric Depression Scale (GDS): recent evidence and development of a shorter version. Clin Gerontol 1986; 5:165–173 26. Almeida OP, Almeida SA: Short versions of the geriatric depression scale: a study of their validity for the diagnosis of a major depressive episode according to ICD-10 and DSM-IV. Int J Geriatr Psychiatry 1999; 14:858–865 27. Brownson RC, Hoehner CM, Day K, et al: Measuring the built environment for physical activity: state of the science. Am J Prev Med 2009; 36:S99–S123.e112 28. Frank LD, Schmid TL, Sallis JF, et al: Linking objectively measured physical activity with objectively measured urban form: findings from SMARTRAQ. Am J Prev Med 2005; 28:117–125 29. Leslie E, Coffee N, Frank L, et al: Walkability of local communities: using geographic information systems to objectively
469
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Built Environment and Depression in Later Life
30.
31.
32.
33.
34.
35. 36.
37.
38. 39.
40.
assess relevant environmental attributes. Health Place 2007; 13:111–122 Frank LD, Andresen MA, Schmid TL: Obesity relationships with community design, physical activity, and time spent in cars. Am J Prev Med 2004; 27:87–96 Ewing R, Schmid T, Killingsworth R, et al: Relationship between urban sprawl and physical activity, obesity, and morbidity. Am J Health Promot 2003; 18:47–57 ABS: Census of Population and Housing—Socio-economic Indexes for Areas. Canberra, Australia, Australian Bureau of Statistics, 2003 Lawton MP, Moss M, Moles E: The suprapersonal neighborhood context of older people: age heterogeneity and well-being. Environ Behav 1984; 16:89–109 Subramanian SV, Kubzansky L, Berkman L, et al: Neighborhood effects on the self-rated health of elders: uncovering the relative importance of structural and service-related neighborhood environments. J Gerontol B Psychol Sci Soc Sci 2006; 61: S153–S160 ABS: 2001 Census of Population and Housing. Canberra, Australia, Australian Bureau of Statistics, 2001 Vink D, Aartsen MJ, Comijs HC, et al: Onset of anxiety and depression in the aging population: comparison of risk factors in a 9-year prospective study. Am J Geriatr Psychiatry 2009; 17:642–652 Oxman TE, Berkman LF, Kasl S, et al: Social support and depressive symptoms in the elderly. Am J Epidemiol 1992; 135: 356–368 Smilkstein: The family APGAR: a proposal for a family function test and its use by physicians. J Fam Pract 1978; 6:1231–1239 Young AF, Russell A, Powers JR: The sense of belonging to a neighbourhood: can it be measured and is it related to health and well being in older women? Soc Sci Med 2004; 59:2627–2637 Charlson ME, Pompei P, Ales KL, et al: A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987; 40:373–383
470
41. Quan H, Sundararajan V, Halfon P, et al: Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data. Med Care 2005; 43:1130–1139 42. Almeida OP, Flicker L, Norman P, et al: Association of cardiovascular risk factors and disease with depression in later life. Am J Geriatr Psychiatry 2007; 15:506–513 43. StataCorp LP: Stata 10.1, TX, StataCorp LP, 2008 44. Mair C, Roux AVD, Galea S: Are neighbourhood characteristics associated with depressive symptoms? A review of evidence. J Epidemiol Community Health 2008; 62:940–946 45. Appleyard D: Livable Streets. Berkeley, CA, University of California Press, 1981 46. Halpern D: Mental Health and the Built Environment: More Than Bricks and Mortar? London, UK, Taylor & Francis, 1995 47. Ellis CD, Lee S-W, Kweon B-S: Retail land use, neighborhood satisfaction and the urban forest: an investigation into the moderating and mediating effects of trees and shrubs. Landsc Urban Plan 2006; 74:70–78 48. Wood L, Frank LD, Giles-Corti B: Sense of community and its relationship with walking and neighborhood design. Soc Sci Med 2010; 70:1381–1390 49. AIHW: Older Australia at a Glance, 4th ed. Canberra, Australian Institute of Health and Welfare, 2007. 50. Choi NG, McDougall GJ: Comparison of depressive symptoms between homebound older adults and ambulatory older adults. Aging Mental Health 2007; 11:310–322 51. Taylor AH, Cable NT, Faulkner G, et al: Physical activity and older adults: a review of health benefits and the effectiveness of interventions. J Sports Sci 2004; 22:703–725 52. Kaufman JS, Cooper RS, McGee DL: Socioeconomic status and health in blacks and whites: the problem of residual confounding and the resiliency of race. Epidemiology 1997; 8:621–628 53. Sugiyama T, Ward Thompson C: Older people’s health, outdoor activity and supportiveness of neighbourhood environments. Landsc Urban Plan 2007; 83:168–175
Am J Geriatr Psychiatry 19:5, May 2011
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.