SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. REFERENCES

Objective

To examine the association between individual- and neighborhood-level disadvantage and self-reported arthritis.

Methods

We used data from a population-based cross-sectional study conducted in 2007 among 10,757 men and women ages 40–65 years, selected from 200 neighborhoods in Brisbane, Queensland, Australia using a stratified 2-stage cluster design. Data were collected using a mail survey (68.5% response). Neighborhood disadvantage was measured using a census-based composite index, and individual disadvantage was measured using self-reported education, household income, and occupation. Arthritis was indicated by self-report. Data were analyzed using multilevel modeling.

Results

The overall rate of self-reported arthritis was 23% (95% confidence interval [95% CI] 22–24). After adjustment for sociodemographic factors, arthritis prevalence was greatest for women (odds ratio [OR] 1.5, 95% CI 1.4–1.7) and in those ages 60–65 years (OR 4.4, 95% CI 3.7–5.2), those with a diploma/associate diploma (OR 1.3, 95% CI 1.1–1.6), those who were permanently unable to work (OR 4.0, 95% CI 3.1–5.3), and those with a household income <$25,999 (OR 2.1, 95% CI 1.7–2.6). Independent of individual-level factors, residents of the most disadvantaged neighborhoods were 42% (OR 1.4, 95% CI 1.2–1.7) more likely than those in the least disadvantaged neighborhoods to self-report arthritis. Cross-level interactions between neighborhood disadvantage and education, occupation, and household income were not significant.

Conclusion

Arthritis prevalence is greater in more socially disadvantaged neighborhoods. These are the first multilevel data to examine the relationship between individual- and neighborhood-level disadvantage upon arthritis and have important implications for policy, health promotion, and other intervention strategies designed to reduce the rates of arthritis, indicating that intervention efforts may need to focus on both people and places.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. REFERENCES

Arthritis, a term encompassing many different types of arthritis, the most often diagnosed being osteoarthritis and rheumatoid arthritis (1), is a common and often debilitating disease. In developed countries it is the most common single cause of disability, impacting both the individual and population levels (2). Australian public health expenditure for arthritis and other musculoskeletal disorders was recently estimated at approximately $4 billion (3), and accounted for the largest proportion of direct health expenditure (31%), amounting to $1.2 billion (3). Individuals with these diseases experience a substantially lower quality of life compared to those without arthritis (1). In addition, and compared to any other disease, arthritis accounts for greater increased dependency to perform activities of daily living (1). Currently there is no known cure for arthritis; however, the promotion of modifiable risk factors to specific at-risk populations offers the potential to reduce arthritis onset, or at least to slow the progression of this disease.

With few exceptions, an inverse relationship exists between social disadvantage and disease (4, 5). Attention is now increasingly being directed toward examining the relationship between socioeconomic status (SES) and arthritis; however, these data still remain sparse. It is well documented that individuals of lower SES, measured by individual parameters such as income, education, and occupation, have lifestyles that are less protective of arthritis, including, but not limited to, greater levels of physical inactivity and obesity (6) and a greater likelihood of smoking (7). Occupation also influences the onset and/or progression of arthritis, whereby individuals with physically demanding occupations such as miners and dockers have a higher prevalence of knee arthritis (8), and physically strenuous jobs that require knee bending, squatting, or heavy lifting significantly increase the risk of knee arthritis (9). Furthermore, data examining area-based parameters of SES show that individuals who reside in areas of greater disadvantage are more likely to participate in less protective lifestyle behaviors that predispose them to arthritis (10).

There is an increasing body of work employing multilevel statistical techniques aimed at disentangling the relative contribution of socioeconomic factors at the individual and neighborhood levels to various diseases, but none to date have examined arthritis. Multilevel analyses have shown that residents of disadvantaged neighborhoods, independent of individual factors, are more likely to be overweight (11) and have lower levels of physical activity (12), both significant modifiable lifestyle-related risk factors that are associated with the onset and progression of arthritis (13, 14). Our primary hypothesis was that there would be differential contributions to arthritis associated with individual- and neighborhood-level disadvantage. A secondary aim was to examine, regardless of individual socioeconomic position (SEP), whether individuals who resided within a more advantaged neighborhood may experience a protective effect against arthritis, or conversely, whether the likelihood of reporting arthritis may be exacerbated if residing in a more disadvantaged neighborhood.

Significance & Innovations

  • These are the first multilevel data to examine the relationship between individual- and neighborhood-level disadvantage upon arthritis.

  • We observed arthritis prevalence to be greater in neighborhoods of increased social disadvantage.

  • We provide novel and important information regarding specific population groups at an increased risk of arthritis and the importance of focusing on both people and places for disease intervention.

PATIENTS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. REFERENCES

This investigation used data from the How Areas in Brisbane Influence Health and Activity study (HABITAT). HABITAT was a multilevel longitudinal (2007–2011) study of physical activity among middle-aged (40–65 years) adults living in Brisbane, Queensland, Australia. The primary focus of HABITAT was to examine patterns of change in physical activity over the period 2007–2011, and to assess the relative contributions of environmental, social, psychological, and sociodemographic factors measured at the area and individual levels to these changes. In the current study, we present findings from the HABITAT baseline survey data, which were collected in May 2007. The HABITAT study was awarded ethical clearance by the Queensland University of Technology Human Research Ethics Committee (ref. 3967H).

Sample design.

Details regarding the HABITAT sampling design have been published elsewhere (15). Briefly, a multistage probability sampling design was used to select a stratified random sample (n = 200) of census collector's districts (CCDs), and from within each CCD, a random sample of people ages 40–65 years (n = 17,000). A CCD is the smallest administrative unit used by the Australian Bureau of Statistics (ABS) to collect census data. In urban areas such as Brisbane, a CCD contains an average of 200 private dwellings that are deemed to be relatively homogeneous in terms of their socioeconomic characteristics (12). CCDs are embedded within a larger suburb; therefore, the area corresponding to and immediately surrounding a CCD is likely to have meaning and significance for their residents. For this reason, we hereafter use the term “neighborhood” to refer to CCDs.

Data collection and response rates.

A structured self-administered questionnaire was developed (16, 17) that asked respondents about their neighborhood; participation in physical activity; correlates of activity, health, and well-being; and sociodemographic characteristics. The questionnaire was administered during May to July 2007 using a mail-survey method (18), and a total of 11,037 usable surveys were returned (response rate 68.5%).

Measures.

Neighborhood disadvantage.

Each of the 200 CCDs was assigned a socioeconomic score using the ABS Index of Relative Socioeconomic Disadvantage (IRSD) (19). The IRSD scores were calculated using 2006 census data and were derived by the ABS using principal components analysis. A CCD IRSD score reflects each area's overall level of disadvantage measured on the basis of 17 variables that capture a wide range of socioeconomic attributes, including education, occupation, income, unemployment, household structure, household tenure, marital status, English language competency, motor vehicle availability, and indigenous status. For analysis, the 200 CCDs were grouped into quintiles based on their IRSD scores, with Q1 denoting the 20% (n = 40) most disadvantaged areas in Brisbane and Q5 denoting the 20% (n = 40) least disadvantaged areas in Brisbane.

Education.

Respondents were asked to provide information about the highest educational qualification completed. A respondent's education was subsequently coded as 1) bachelor degree or higher (the latter included postgraduate diploma, master's degree, or doctorate), 2) diploma (associate or undergraduate), 3) vocational (trade or business certificate or apprenticeship), and 4) no postsecondary school qualifications.

Occupation.

Respondents who were employed at the time of completing the survey were asked to indicate their job title and then to describe the main tasks or duties they performed. This information was subsequently coded to the Australian Standard Classification of Occupations (ASCO) (20). For the purposes of this study, the original 9-level ASCO classification was recoded into 3 categories: 1) managers/professionals (managers and administrators, professionals, and paraprofessionals), 2) white-collar employees (clerks, salespersons, and personal service workers), and 3) blue-collar employees (tradespersons, plant and machine operators and drivers, and laborers and related workers). Respondents who were not employed were categorized as follows: 4) home duties, 5) retired, 6) permanently unable to work, 7) other (not easily classifiable), and 8) missing.

Income.

Respondents were asked to indicate their total annual household income (including pensions, allowances, and investments) using a 14-category measure that was subsequently recoded into 6 groups for analysis: 1) $130,000 (Australian dollars) or more, 2) $72,800–129,999, 3) $41,600–72,799, 4) $26,000–41,599, 5) less than $25,999, and 6) missing.

Self-reported arthritis.

Self-reported arthritis was measured using responses to a question that asked, “Have you ever been told by a doctor or nurse that you have any of the LONG-TERM health conditions listed below? (please only include those conditions that have lasted, or are likely to last, for six (6) months or more).” Arthritis was 1 of 8 conditions listed, and respondents were asked to indicate “yes” (coded 1) or “no” (coded 0) for each condition. This question has been used extensively in previous Australian health research (21).

Statistical analysis.

Of the 11,037 returned surveys, a small number had missing data for education (n = 47 [0.43%]), and a larger number were missing self-reported arthritis status (n = 233 [2.1%]). By contrast, 916 respondents (8.3%) had missing data for occupation and 1,561 (14.1%) had missing data on income. Given the complexity of the statistical analysis and the large number of variables that were included in the multilevel models, it was deemed inappropriate to include data cells that contained small numbers, as this would have increased the likelihood of model convergence problems. Also, the exclusion of so few cases for education and arthritis (i.e., 280 or 2.5% in total) would not have affected the study's findings or our interpretation of them. After exclusion of missing data, this reduced the analytic sample to 10,757 (97.5% of the total sample).

Multilevel logistic modeling was used to assess whether neighborhood disadvantage and individual- and household-level SEP were associated with arthritis. These relationships were analyzed with MLwiN, version 2.22 (22), using a binomial logit-link model with the predictive-penalized quasi-likelihood procedure and second-order linearization using the iterative generalized least-squares algorithm (23). The analyses were conducted in 3 stages. First, we specified a null model (model 1) that comprised individuals (level 1) nested in neighborhoods (level 2) with no individual- or area-level variables in the fixed part of the model. Substantive interest for the null model focuses on the neighborhood-level random term, which if significant (indicated using Wald's chi-square test), suggests between-neighborhood variation in reported arthritis. Second, the null model was extended to include individual- and household-level fixed effects for sex, age, education, occupation, and household income (model 2) and then neighborhood disadvantage (model 3). Results for models 1–3 are reported as odds ratios (ORs) and 95% confidence intervals (95% CIs). Third, cross-level interactions were assessed by including interaction terms that reflected the impact of different combinations of individual- and household-level SEP and neighborhood disadvantage on the likelihood of reporting arthritis. The substantive focus of the interaction analyses is on whether the associations between education, occupation, household income, and arthritis differed depending on the extent of neighborhood disadvantage. The fit of the interaction models was assessed using a joint Wald's chi-square test.

Finally, as part of our preliminary multilevel analyses, we tested for possible interactions between sex and SEP and found no evidence that the association between education (P = 0.560), occupation (P = 0.162), household income (P = 0.403), neighborhood disadvantage (P = 0.604), and arthritis differed for men and women. Consequently, we present our results for men and women combined.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. REFERENCES

Table 1 shows the proportion of respondents who reported experiencing arthritis as a chronic condition. The overall rate of reported arthritis was 23.3% (95% CI 22.4–24.1), and the rates were highest for women (27.1%; 95% CI 26.0–28.3), those ages 60–65 years (40.0%; 95% CI 37.8–42.2), the least educated (28.2%; 95% CI 26.8–29.5), those who were permanently unable to work (58.8%; 95% CI 53.3–64.3), respondents from low-income families (42.2%; 95% CI 39.1–45.3), and residents of the most disadvantaged neighborhoods (31.4%; 95% CI 29.1–33.7).

Table 1. Number of respondents in the study sample and the proportion (95% CI) who reported experiencing arthritis as a long-term chronic condition*
 N% reporting arthritis (95% CI)
  • *

    95% CI = 95% confidence interval.

All persons10,75723.3 (22.4–24.1)
Sex  
 Male4,75218.4 (17.3–19.5)
 Female6,00527.1 (26.0–28.3)
Age, years  
 40–442,20710.3 (9.0–11.6)
 45–492,38816.1 (14.6–17.6)
 50–542,20723.1 (21.3–24.8)
 55–592,03730.2 (28.2–32.2)
 60–651,91840.0 (37.8–42.2)
Educational attainment  
 Bachelor's degree or higher3,39117.0 (15.8–18.3)
 Diploma/associate diploma1,25123.3 (20.9–25.6)
 Certificate (trade/business)1,91423.6 (21.7–25.5)
 Secondary school4,20128.2 (26.8–29.5)
Occupation  
 Manager/professional3,61016.7 (15.4–17.9)
 White collar2,36523.2 (21.5–24.9)
 Blue collar1,52020.6 (18.6–22.6)
 Home duties66727.6 (24.1–31.0)
 Retired94241.1 (37.9–44.2)
 Permanently unable to work31158.8 (53.3–64.3)
 Other (not easily classifiable)42626.5 (22.3–30.7)
 Missing91619.0 (16.5–21.5)
Household income  
 ≥$130,0001,86113.3 (11.7–14.8)
 $72,800–129,9992,79518.2 (16.7–19.6)
 $41,600–72,7992,38724.0 (22.3–25.7)
 $26,000–41,5991,16530.6 (28.0–33.3)
 <$25,99998842.2 (39.1–45.3)
 Missing1,56125.7 (23.5–27.9)
Neighborhood disadvantage  
 Quintile 5 (least  disadvantaged)2,51518.5 (16.9–20.0)
 Quintile 42,65421.1 (19.5–22.7)
 Quintile 32,21524.6 (22.8–26.4)
 Quintile 21,79524.4 (22.4–26.4)
 Quintile 1 (most  disadvantaged)1,57831.4 (29.1–33.7)

Table 2 shows the multilevel association between individual- and household-level sociodemographic factors and neighborhood disadvantage and the likelihood of reporting arthritis. The results for the null model (model 1) show that the probability of reporting arthritis differed significantly across the neighborhoods (P ≤ 0.001). The differences were successively reduced with adjustment for the individual- and household-level factors (by 79.2% for model 2 relative to the null model) and neighborhood disadvantage (92.2%), after which there was no longer between-neighborhood variation in self-reported arthritis (P = 0.204). Similarly, in the fully adjusted model (model 3), there was no significant between-neighborhood variation in the probability of reporting arthritis (P = 0.602).

Table 2. Individual- and neighborhood-level effects on self-reported arthritis (n = 10,757)*
 Model 1Model 2Model 3§
  • *

    Values are the odds ratio (95% confidence interval) unless otherwise indicated.

  • Model 1 (null model): between-neighborhood variation in arthritis unconditioned on any other factor.

  • Model 2: model 1 plus adjustment for within-neighborhood variation in sex, age, household income, education, and occupation.

  • §

    Model 3: model 2 plus adjustment for neighborhood socioeconomic disadvantage.

  • Least disadvantaged quintile.

  • #

    Variance estimate (standard error).

  • **

    P value for a joint Wald's chi-square test.

Fixed effects   
 Sex   
  Male (referent) 1.001.00
  Female 1.50 (1.35–1.67)1.51 (1.36–1.68)
  P ≤ 0.001≤ 0.001
 Age, years   
  40–44 (referent) 1.001.00
  45–49 1.62 (1.35–1.93)1.62 (1.36–1.94)
  50–54 2.49 (2.09–2.96)2.50 (2.10–2.97)
  55–59 3.33 (2.80–3.95)3.36 (2.83–3.99)
  60–65 4.38 (3.66–5.23)4.42 (3.70–5.28)
  P ≤ 0.001≤ 0.001
 Educational attainment   
  Bachelor's degree or higher (referent) 1.001.00
  Diploma/associate diploma 1.31 (1.11–1.55)1.30 (1.10–1.54)
  Certificate (trade/business) 1.29 (1.10–1.51)1.26 (1.08–1.47)
  Secondary school 1.29 (1.13–1.47)1.25 (1.10–1.44)
  P ≤ 0.0010.002
 Occupation   
  Manager/professional (referent) 1.001.00
  White collar 1.01 (0.87–1.17)1.01 (0.87–1.17)
  Blue collar 1.02 (0.86–1.22)1.00 (0.84–1.19)
  Home duties 1.13 (0.91–1.39)1.12 (0.91–1.39)
  Retired 1.29 (1.07–1.56)1.29 (1.07–1.56)
  Permanently unable to work 4.02 (3.05–5.29)3.85 (2.93–5.07)
  Other (not easily classifiable) 1.22 (0.95–1.56)1.20 (0.94–1.54)
  P ≤ 0.001≤ 0.001
 Household income   
  ≥$130,000 (referent) 1.001.00
  $72,800–129,999 1.32 (1.11–1.57)1.28 (1.08–1.53)
  $41,600–72,799 1.59 (1.33–1.89)1.51 (1.26–1.80)
  $26,000–41,599 1.86 (1.53–2.27)1.75 (1.43–2.14)
  <$25,999 2.14 (1.73–2.64)1.96 (1.58–2.43)
  P ≤ 0.001≤ 0.001
 Neighborhood disadvantage   
  Quintile 5 (referent)  1.00
  Quintile 4  1.09 (0.94–1.26)
  Quintile 3  1.23 (1.05–1.43)
  Quintile 2  1.17 (0.99–1.38)
  Quintile 1  1.42 (1.20–1.68)
  P  ≤ 0.001
Random effects#0.077 (0.019)0.016 (0.012)0.006 (0.011)
 P**≤ 0.0010.2040.602

After simultaneously adjusting for all sociodemographic factors (model 2), statistically significant associations (P ≤ 0.001) were observed between self-reported arthritis and sex, age, education, occupation, and household income, with rates being highest for women (OR 1.50, 95% CI 1.35–1.67), those ages 60–65 years (OR 4.38, 95% CI 3.66–5.23), respondents with a diploma/associate diploma (OR 1.31, 95% CI 1.11–1.55), those permanently unable to work (OR 4.02, 95% CI 3.05–5.29), and those living in households where the annual income was less than $25,999 (OR 2.14, 95% CI 1.73–2.64). When neighborhood disadvantage was added to the model (model 3), associations between each of the sociodemographic factors and arthritis remained largely unchanged or were slightly attenuated. Independent of sex, age, education, occupation, and household income, residents of the most disadvantaged neighborhoods were 42% (OR 1.42, 95% CI 1.20–1.68) more likely than their counterparts in the least disadvantaged neighborhoods to report that they experienced arthritis as a chronic condition.

Tests of cross-level interactions were not statistically significant between neighborhood disadvantage and education (P = 0.164), occupation (P = 0.628), and household income (P = 0.389) (Figure 1).

thumbnail image

Figure 1. Cross-level interactions between A, education, B, occupation, C, household income, and neighborhood disadvantage and the probability of reporting arthritis in men and women ages 40–65 years in 2007. Models include age, sex, education, occupation, household income, and neighborhood disadvantage. Dis. = disadvantaged; Dip/Assoc = diploma/associate diploma; Mgr/Prof = manager/professional.

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. REFERENCES

This study shows novel insights into the social patterning of arthritis. Individuals from disadvantaged neighborhoods were more likely to report arthritis compared to individuals residing in less disadvantaged neighborhoods, independent of education, occupation, and household income. Patterns of association also suggested that individual- and neighborhood-level socioeconomic factors may combine to either amplify the susceptibility to disadvantage or offer additional protection against arthritis. In light of a paucity of data, our current understanding of why advantaged and disadvantaged neighborhoods differ in their rates of self-reported arthritis is limited; therefore, we provide a speculative discussion to inform future research.

When examining the role played by social disadvantage in health outcomes, we provide evidence for biologic plausibility, a criterion clearly demonstrated within the widely documented social gradient of health (4). Biologic plausibility is also suggested as being inherent in the role played by social capital, a complex construct concerned with “features of social organization such as trust, norms, and networks that can improve the efficiency of society by facilitating coordinated actions” (24) upon health outcomes. Social capital may contribute to differences in rates of arthritis between advantaged and disadvantaged neighborhoods, potentially resulting from 3 processes of 1) concomitant differences in mutual concern for others, 2) the extent of social networks, and 3) their normative environments. Preventive health care and increased participation in medical checkups may also be a characteristic of a neighborhood with greater social capital, thereby enabling residents to identify potential health problems prior to onset. Social capital has the potential to reduce individuals' focus on disease and illness, potentially explaining why individuals from disadvantaged neighborhoods are more likely to self-rate their general health as poor, compared to their counterparts residing in less disadvantaged areas (25). Neighborhoods of greater social capital share “norms” that influence preventive health behaviors such as physical activity and lower caloric intake, both of which are key predictors of body habitus and obesity. Furthermore, neighborhoods with greater social capital potentially have greater involvement with social networks, community activities, and social engagement (26). Individuals in those neighborhoods may be more able to secure health-promoting resources (27, 28), which may include safe walking paths with good lighting and even surfaces, and this may reduce the likelihood of traumatic injury to the knees or hips related to falls, high-impact forces, or malalignment of the joint during uneven or altered biomechanical loading (29). Trauma to joints increases the likelihood of developing bone marrow lesions, a factor well documented as being associated with pain in those with osteoarthritis (30).

Positive mental well-being is a fundamental element that enables individuals to cope with adversity, operationalized through factors such as resilience and positive adaptation (31). By residing in a socially disadvantaged area, the individual not only experiences their individual-level dimension of inequality, but may also perceive an unequal distribution of economic and social resources compared to other areas, potentially resulting in an internalization of the emotional and cognitive effects of insecurity at many levels (31). Indeed, the importance of social and psychological dimensions of deprivation at multiple levels is gaining increased international recognition due to efforts to address poorer health outcomes experienced by those of greater social disadvantage (4). Considering that resilience and positive adaptation are likely to act as protective factors with regard to symptoms associated with poorer health in general, positive mental well-being may improve an individual's ability to cope with pain, resulting in an improved quality of life. However, pain is but one characteristic of arthritis, with radiologic diagnosis being another. Should the individual be less affected by pain, there is less likelihood that a medical practitioner may determine a need for radiology services, reducing the likelihood of diagnosis.

Obesity is the strongest modifiable predictor of arthritis onset and progression, especially knee osteoarthritis (13), and increases the likelihood of reduced cartilage volume (32), the hallmark characteristic of arthritis. Given that physical activity is an important predictor of obesity, this is worthy of discussion in the context of social determinants. The built environment plays a key role in terms of influencing physical activity in neighborhoods, with recent data showing a greater propensity for individuals residing in less disadvantaged areas to participate in more physically active lifestyles (12). However, given that our current study did not examine the role of physical activity on arthritis, we are only able at this point to speculate about the complexity between disadvantage measured at the neighborhood and individual levels and arthritis. For instance, undertaking a higher-skilled occupation may likely result in lower levels of work-related physical activity, and the opposite could be suggested of lower-skilled occupations; however, individuals who have a physically demanding occupation may not recognize or report those physical demands in a self-administered survey. Clearly, this area of inquiry begs immediate attention to elucidate by use of multilevel analyses the role played by physical activity in the onset and progression of arthritis.

Income and education are inextricably linked, with one influencing the other, and it is well documented that individuals with greater education and/or income are more likely to be health literate compared to individuals with less education and/or income (33). Given that education and income reflect material and intellectual resources, there is an inherent assumption of an important dose-response relationship between those individual measures of SES and improved health literacy (33). Taken in context, social capital and health literacy skills combined may explain the nonsignificant patterns of association we observed that suggested a potential protective effect of residing in a less disadvantaged neighborhood for individuals with lower educational attainment or household income.

We observed a greater likelihood of self-reported arthritis in women compared to men, which is indicative of the well-documented sex bias observed for this disease. Although the mechanisms by which female sex increases the risk of arthritis are unknown, they may be related to a combination of hormonal, heritability, and/or body habitus factors (13). However, differential variation in arthritis between the sexes when accounting for social disadvantage may also be partly explained by a potential dose-response aspect of social capital, whereby the well-documented dual workload of women (34) may limit the amount of time spent interacting with their immediate neighborhood, and thus results in an increased likelihood of reporting ill health (34).

The increased risk of arthritis in association with age is not surprising, given that this is a well-documented predisposing risk factor for this disease (35). Furthermore, the observed association between arthritis and being permanently unable to work is also not surprising, given that arthritis is associated with reduced mobility and increased difficulty in performing acts of daily living (1, 3). We are unable to comment further on this observed association due to the cross-sectional nature of our analyses.

This study has some limitations. First, the categorization of occupations may have limited our ability to determine an association with arthritis due to the inability to assign occupation type to 26.5% of participants, including those who were volunteers, held numerous part-time positions, were retired, were current students, or were home based, and the 19% for whom data were missing. Furthermore, and as previously reported, the areal units employed for analysis within the HABITAT cohort were selected based on sampling convenience (36), and not reasons specifically related to arthritis prevalence; given this, we hypothesize that the potential impact of living in a disadvantaged neighborhood on the prevalence of self-reported arthritis would be underestimated. Nonresponse for the baseline HABITAT study was 31.5%, with greater nonresponse from individuals in more disadvantaged neighborhoods (12). The population was identified from the Commonwealth Electoral Roll, which has been reported as underrepresenting individuals who are disadvantaged, transient or homeless, and migrant (15). Given this, and the fact that lifestyle behaviors associated with arthritis are more likely to be observed in individuals of lower SES (10), it is possible that the relationship we report between neighborhood disadvantage and arthritis may be underestimated. However, we also recognize that the OR may overestimate the relative risks of arthritis (37). Our findings are based on a research design that achieved a moderate individual-level response rate of 68.5%, and a response rate that followed an inverse association across the deciles of neighborhood disadvantage. We therefore need to consider the likely bias attributable to nonresponse and how this might affect this study's inferences to the wider population. Previous studies show that persons from socioeconomically disadvantaged backgrounds (38) and residents of more deprived neighborhoods (39) are the least likely to respond to or participate in survey research. As a result, population-based samples typically underrepresent the most disadvantaged and overrepresent the advantaged, the likely consequence of which is a socioeconomically truncated sample resulting in an underestimation of the magnitude of socioeconomic variability in self-reported arthritis. Therefore, the neighborhood- and individual-level socioeconomic differences in arthritis shown in this study, although significant, may be an underestimate of the “true” magnitude of socioeconomic differences in the population. Also, our finding of an association between neighborhood disadvantage and self-reported arthritis might be confounded by individual-level socioeconomic factors not included in the models. However, we included 3 of the most widely used indicators of an individual's socioeconomic characteristics (education, occupation, and income), and given the correlation among these measures (40), it is likely that these socioeconomic indicators are capturing some of the unmeasured influences of other socioeconomic factors not included in the models. Alternatively, it may be that the inclusion of individual-level measures of SES resulted in “overadjustment,” which provides evidence for the possibility of an even stronger contextual effect on self-reported arthritis than was observed in this study. If education, occupation, and household income represent part of the pathway via which neighborhood disadvantage influences the likelihood of experiencing arthritis, then simultaneously modeling individual-level socioeconomic variables may have inappropriately attenuated the variation that was more correctly attributable to neighborhood disadvantage (41). Finally, we were unable to determine the proportions of osteoarthritis and rheumatoid arthritis within our measure of arthritis due to the method of disease identification; however, of the overall arthritis prevalence, we may expect to observe ∼8% of the former and ∼2% of the latter (1). In the absence of published data examining this question and in light of the current limited understanding examining social disadvantage and arthritis, our findings encourage further research into this area of inquiry.

We conclude that arthritis prevalence is greater in neighborhoods of increased social disadvantage. This study is the first to examine the association between social disadvantage and arthritis using a multilevel analysis, and therefore provides novel and important information regarding specific population groups at an increased risk of arthritis. These data also suggest the importance of focusing on both people and places for disease intervention. Without this information, we are limited in our ability to intervene to reduce the prevalence of arthritis, and to identify potential target groups for preventive health programs. Given that arthritis is not yet curable and little is currently known about the social determinants of arthritis diseases, our study prompts further research into examining this association, and potentially encourages increased attention from health practitioners toward clients from socially disadvantaged groups and neighborhoods.

AUTHOR CONTRIBUTIONS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. REFERENCES

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Brennan had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Brennan, Turrell.

Acquisition of data. Turrell.

Analysis and interpretation of data. Brennan, Turrell.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. REFERENCES
  • 1
    Australian Institute of Health and Welfare. Snapshot of arthritis in Australia 2010. Canberra: AIHW; 2010.
  • 2
    Pendleton A, Arden N, Dougados M, Doherty M, Bannwarth B, Bijlsma JW, et al. EULAR recommendations for the management of knee osteoarthritis: report of a Task Force of the Standing Committee for International Clinical Studies Including Therapeutic Trials (ESCISIT). Ann Rheum Dis 2000; 59: 93644.
  • 3
    Australian Institute for Health and Welfare. Arthritis and osteoporosis in Australia 2008. Canberra: AIHW, National Centre for Monitoring Arthritis and Other Musculoskeletal Conditions; 2008.
  • 4
    Wilkinson RG, Marmot MG. Social determinants of health: the solid facts. Copenhagen: World Health Organization European Region; 1998.
  • 5
    Turrell G, Stanley L, de Looper M, Oldenburg B. Health inequalities in Australia: morbidity, health behaviours, risk factors and health service use. Canberra: Queensland University of Technology, Australian Institute of Health and Welfare; 2006.
  • 6
    Ball K, Mishra G, Crawford D. Which aspects of socioeconomic status are related to obesity among men and women? Int J Obes 2002; 26: 55965.
  • 7
    Gilman SE. Educational attainment and cigarette smoking: a causal association? Int J Epidemiol 2008; 37: 61524.
  • 8
    Partridge R, Duthie JJ. Rheumatism in dockers and civil servants: a comparison of heavy manual and sedentary workers. Ann Rheum Dis 1968; 27: 55969.
  • 9
    Felson D, Hannan MT, Naimark A, Berkeley J, Gordon G, Wilson PW, et al. Occupational physical demands, knee bending, and knee osteoarthritis: results from the Framingham Study. J Rheumatol 1991; 18: 158792.
  • 10
    Brennan SL, Henry MJ, Nicholson GC, Kotowicz MA, Pasco JA. Socioeconomic status and risk factors for obesity and metabolic disorders in a population-based sample of adult females. Prev Med 2009; 49: 16571.
  • 11
    King T, Kavanagh AM, Jolley D, Turrell G, Crawford D. Weight and place: a multilevel cross-sectional survey of area-level social disadvantage and over-weight/obesity in Australia. Int J Obes 2006; 30: 2817.
  • 12
    Turrell G, Haynes M, Burton NW, Giles-Corti B, Oldenburg B, Wilson LA, et al. Neighborhood disadvantage and physical activity: baseline results from the HABITAT multilevel longitudinal study. Ann Epidemiol 2010; 20: 17181.
  • 13
    Felson D. Obesity and osteoarthritis of the knee. Bull Rheum Dis 1992; 41: 67.
  • 14
    Racunica TL, Teichtahl AJ, Wang Y, Wluka AE, English DR, Giles GG, et al. Effect of physical activity on articular knee joint structures in community-based adults. Arthritis Rheum 2007; 57: 12618.
  • 15
    Burton NW, Haynes M, Wilson LA, Giles-Corti B, Oldenburg B, Brown WJ, et al. HABITAT: a longitudinal multilevel study of physical activity change in mid-aged adults. BMC Public Health 2009; 9: 76.
  • 16
    Burton NW, Turrell G, Oldenburg B, Sallis J. The relative contribution of psychological, social, and environmental variables to physical activity. J Phys Act Health 2005; 2: 18196.
  • 17
    Burton NW, Oldenburg B, Sallis JF, Turrell G. Measuring psychological, social, and environmental influences on leisure-time physical activity among adults. Aust N Z Public Health 2007; 31: 3643.
  • 18
    Dillman DA. Mail and internet surveys: the tailored design method: Chichester (UK): Wiley; 2000.
  • 19
    Australian Bureau of Statistics. Socio-economic indices for areas (SEIFA): technical paper. Canberra: ABS; 2006.
  • 20
    Australian Bureau of Statistics. Australian Standard Classification of Occupations. 2nd ed. Canberra: ABS; 1997.
  • 21
    Australian Bureau of Statistics. 2007-08 national health survey: summary of results. Canberra: ABS; 2009.
  • 22
    Rasbash J, Steele F, Browne W, Prosser B. A user's guide to MLwiN (version 2): documentation version 2.1e. University of Bristol: Bristol; 2003.
  • 23
    Goldstein H. Multilevel statistical models. London: Arnold; 2003.
  • 24
    Putnam RD. Making democracy work: civic traditions in modern Italy. Princeton (NJ): Princeton University Press; 1993.
  • 25
    Stafford M, Marmot M. Neighbourhood deprivation and health: does it affect us all equally? Int J Epidemiol 2003; 32: 35766.
  • 26
    Kawachi I, Berkman L. Social cohesion, social capital, and health. In: Kawachi I, Berkman L, editors. Social epidemiology. New York: Oxford University Press; 2000. p. 17490.
  • 27
    Subramanian SV, Lochner KA, Kawachi I. Neighbourhood differences in social capital: a compositional artifact or a contextual construct? Health Place 2003; 9: 3344.
  • 28
    Turrell G, Kavanagh A, Subramanian SV. Area variation in mortality in Tasmania (Australia): the contributions of socioeconomic disadvantage, social capital and geographic remoteness. Health Place 2006; 12: 291305.
  • 29
    Felson D, McLaughlin S, Goggins J, LaValley MP, Gale E, Totterman S, et al. Bone marrow edema and its relation to progression of knee osteoarthritis. Ann Intern Med 2003; 139: 3306.
  • 30
    Felson DT, Chaisson CE, Hill CL, Totterman SM, Gale ME, Skinner KM, et al. The association of bone marrow lesions with pain in knee osteoarthritis. Ann Intern Med 2001; 134: 5419.
  • 31
    Friedli L, for the World Health Organization Regional Office for Europe. Mental health, resilience and inequalities. Denmark: World Health Organization; 2009.
  • 32
    Ding C, Parameswaran V, Cicuttini FM, Burgess J, Zhai G, Quinn S, et al. Association between leptin, body composition, sex and knee cartilage morphology in older adults: the Tasmanian older adult cohort (TASOAC) study. Ann Rheum Dis 2008; 67: 125661.
  • 33
    Kelleher J. Cultural literacy and health. Epidemiol 2002; 13: 497500.
  • 34
    Krantz G, Berntsson L, Lundberg U. Total workload, work stress and perceived symptoms in Swedish male and female white-collar employees. Eur J Public Health 2005; 15: 20914.
  • 35
    Felson DT, Naimark A, Anderson J, Kazis L, Castelli W, Meenan RF. The prevalence of knee osteoarthritis in the elderly: the Framingham Osteoarthritis Study. Arthritis Rheum 1987; 30: 9148.
  • 36
    Turrell G, Bentley R, Thomas T, Jolley D, Subramanian S, Kavanagh AM. A multilevel study of area socio-economic status and food purchasing behaviour. Public Health Nutr 2009; 12: 207483.
  • 37
    Barros AJ, Hirakata VN. Alternatives for logistic regression in cross-sectional studies: an empirical comparison of models that directly estimate the prevalence ratio. BMC Med Res Methodol 2003; 3: 21.
  • 38
    Turrell G, Patterson C, Oldenburg B, Gould T, Roy MA. The socio-economic patterning of survey participation and non-response error in a multilevel study of food purchasing behaviour: area- and individual-level characteristics. Public Health Nutr 2003; 6: 1819.
  • 39
    Kavanagh AM, Goller JL, King T, Jolley D, Crawford D, Turrell G. Urban area disadvantage and physical activity: a multilevel study in Melbourne, Australia. J Epidemiol Comm Health 2005; 59: 93440.
  • 40
    Turrell G, Hewitt B, Patterson C, Oldenburg B. Measuring socio-economic position in dietary research: is choice of socio-economic indicator important? Public Health Nutr 2003; 6: 191200.
  • 41
    Diez-Roux AV. Estimating neighborhood health effects: the challenges of causal inference in a complex world. Soc Sci Med 2004; 58: 195360.