Independent Influences of Current and Childhood Socioeconomic Status on Health Outcomes in a North Carolina Family Practice Sample of Arthritis Patients

Authors

  • Antoine R. Baldassari,

    1. University of North Carolina at Chapel Hill
    Search for more papers by this author
  • Rebecca J. Cleveland,

    1. University of North Carolina at Chapel Hill
    Search for more papers by this author
  • Leigh F. Callahan

    Corresponding author
    1. University of North Carolina at Chapel Hill
    • University of North Carolina at Chapel Hill, 3300 Thurston Building, CB #7280, Chapel Hill, NC 27599. E-mail: Leigh_Callahan@med.unc.edu

    Search for more papers by this author
    • Dr. Callahan has received consultant fees, speaking fees, and/or honoraria (less than $10,000 each) from the NIH, the Canadian Arthritis Network, the University of Washington Pain Behaviors Study, and the National Association of Chronic Disease Directors.


Abstract

Objective

Compelling evidence suggests that socioeconomic status (SES) is a determinant of health outcomes among persons with arthritis. SES in early life has likewise been associated with various aspects of health, but the connection between childhood SES and health among people with arthritis remains to be investigated. The purpose of this study was to determine the influences of current and childhood SES on self-reported disability, depression, and physical and mental health among people with self-reported doctor-diagnosed arthritis.

Methods

Data originated from a North Carolina network of primary care centers. Participants with self-reported arthritis with complete sociodemographic and relevant health information were retained in our sample (n = 782). We created summary measures for current and childhood SES from indicators of education, occupation, and homeownership, using parental SES as a proxy for participants' childhood SES. Linear regression models were used to assess the associations between health outcomes and SES variables separately and together, adjusting for key covariates.

Results

Lower childhood and current SES scores were associated with worse disability and physical health. Current SES was furthermore associated with mental health and depressive symptoms. Associations of low current and childhood SES with health outcomes remained significant when concurrently included in a linear model.

Conclusion

Childhood and current SES are both determinants of health among persons with arthritis. This underscores the importance of childhood SES as a determinant of adult health among individuals with arthritis. Further studies should focus on these associations in different populations and across different types of arthritis.

INTRODUCTION

Socioeconomic status (SES) is a recognized determinant of health. The association is well documented in arthritis, as socioeconomic markers such as occupation and educational attainments have been convincingly tied to the health of individuals with arthritis diagnoses, including rheumatoid arthritis (RA), osteoarthritis (OA), lupus, and fibromyalgia ([1-4]).

Studies have placed increasing emphasis on socioeconomic conditions throughout the life course, and mounting evidence suggests that SES during childhood lastingly influences mortality ([5]) and various dimensions of health in adulthood ([6-9]) independently of later socioeconomic circumstances. Likewise, childhood SES has been associated with etiology and health outcomes across a range of chronic conditions ([10-12]). Although tremendous advances have been made toward assessing the influence of childhood socioeconomic circumstances in cardiovascular disease ([10]), the literature remains at an early stage with regard to chronic musculoskeletal disorders. Arthritis studies have so far focused on early socioeconomic determinants of RA development; to date, blue-collar paternal occupation, low maternal education, and low childhood SES sustained into adulthood were each associated with a greater vulnerability to RA ([13-15]), whereas self-assessed social class was not found to affect RA prevalence ([16]). Additionally, one study found arthritis morbidity to be higher among individuals reared in families with low SES ([17]).

Although the etiologic relevance of childhood SES in arthritis has received increasing attention, the relationship between the early socioeconomic environment and health outcomes among individuals with arthritis has yet to be investigated. This study aimed to explore independent associations of adult and childhood SES with self-reported disability, depression, physical health, and mental health in a cohort of North Carolinians with self-reported doctor-diagnosed arthritis.

Box 1. Significance & Innovations

  • Current socioeconomic status (SES) has consistently been associated with worse health outcomes among people with arthritis, but there has yet to be a study of the connection between early SES and health outcomes within that population.
  • We observed significant associations between low childhood SES and self-reported disability and physical health in our sample of participants with self-reported arthritis. Current SES was likewise tied to these health outcome measures, as well as to mental health and depressive symptoms.
  • These associations remained significant when both SES measures were included in the same linear models, implying that childhood and current SES may independently influence health outcomes.

MATERIALS AND METHODS

Study design.

The North Carolina Family Medicine Research Network (NC-FM-RN) is a statewide practice-based network of 22 family medical practices selected to represent the geographic and racial/ethnic diversity of North Carolina. Over the course of 4 weeks in 2001, individuals ages >18 years were offered to be recruited into the NC-FM-RN cohort during their visit to a participating practice. Consenting participants were given a questionnaire inquiring about demographics, common diseases, risk factors, and health habits. The cohort was periodically enriched with new participants (2004, 2005, 2008), and is described in greater detail in previous studies ([18]).

Data originated from complementary phone surveys administered to eligible NC-FM-RN respondents (n = 4,442) in 2004 and 2006 within the Individual and Community Social Determinants of Arthritis Outcomes (SODE) study. The first survey focused on current demographics, health status, attitudes and beliefs, chronic health conditions, and perceptions of neighborhood environment (n = 2,479), and the second survey extended to parental characteristics and childhood circumstances (n = 1,541). Our study focused on the 782 SODE participants who reported having doctor-diagnosed arthritis, defined according to the 2002 arthritis module of the Behavior Risk Factor Surveillance System ([17]), and provided all relevant sociodemographic and health information. The flow of participants from the NC-FM-RN to the current study is detailed in Figure 1. All study materials and methods were approved by the University of North Carolina at Chapel Hill Biomedical Institutional Review Board.

Figure 1.

Flow chart of participants used in our study. NC-FM-RN = North Carolina Family Medicine Research Network; SODE = Individual and Community Social Determinants of Arthritis Outcomes Study; T1 = time 1; T2 = time 2; BMI = body mass index; HAQ = Health Assessment Questionnaire; CES-D = Center for Epidemiologic Studies Depression Scale; PCS = physical component summary (from Short Form 12 version 2 [SF-12v2]); MCS = mental component summary (from SF-12v2).

Measures.

Health outcomes.

Disability was evaluated using the Stanford Health Assessment Questionnaire (HAQ) 20-item disability scale, a widely used self-reported measure of disability in which respondents rate the extent to which they can perform a set of common tasks, scored on an integer scale ranging from 0–3 (where 0 = without any difficulty, 1 = with some difficulty, 2 = with much difficulty, and 3 = unable to do). Activities are organized across 8 categories (dressing and grooming, arising, eating, walking, hygiene, reach, grip, and activities) and a mean score out of 3 is produced for each category from corresponding abilities, with higher scores indicating greater disability ([19]).

We used the 12-item Short Form version 2 (SF-12v2) survey instrument, an abbreviated version of the 36-item Short Form health survey, to rate respondent health. Two distinct scores assessing physical and mental health are created for each individual: the physical component summary (PCS) score and the mental component summary (MCS) score, respectively. The MCS and PCS scores range from 0–100, where higher scores indicate better health. The SF-12v2 has been shown to be a reliable scale in general populations ([20]).

Depressive symptoms were rated using the Center for Epidemiologic Studies Depression Scale (CES-D). Respondents are asked to assess some of their behaviors, feelings, and outlooks (e.g., “I felt depressed,” “I talked less than usual”) by indicating the number of times they experienced each of the 20 symptoms over the past week, ranging from rarely or none of the time (<1 day) to most or all of the time (5–7 days). An integer score out of 3 is created for each symptom, with higher scores indicating greater symptom severity, and the 20 scores are summed to produce a participant's CES-D score out of 60. The CES-D has been shown to be a reliable measure of depressive symptoms in general populations ([21]).

All sample participants had a HAQ score, whereas 36 (4.6%) lacked a CES-D score and 12 (1.5%) did not have a score on the SF-12v2 measures of health (Figure 1).

SES variables.

It has been suggested that maternal education, paternal occupation, and the financial situation of the family adequately reflect SES in childhood as it relates to health ([22]). In the absence of data on parental income or accumulated assets, we used parental homeownership at the time of childhood as an indicator of financial wealth, considering the role of homeownership as the foremost wealth management avenue in the US. We therefore assessed childhood SES by using participant-provided information on their mother's highest level of education, their father's occupation, and their parents' homeownership status during childhood. Parental homeownership was determined by asking participants to describe living arrangements in the household in which they were raised, and occupations were categorized as professional or nonprofessional according to 2000 US Census occupational codes; we used educational or occupational information on the designated primary caretaker for participants with missing data on the mother or father (n = 26 [3.3%] and n = 36 [4.6%], respectively).

Participants indicated their own level of education, their occupation, and whether they were homeowners. These characteristics are frequently used in epidemiologic studies concerned with evaluating current SES. Consequently, current and childhood SES were identically assessed by accounting for a unique set of socioeconomic circumstances at 2 points in the life course.

Covariates.

The covariates used in all of our analyses were known contributors to self-reported health outcomes and consisted of body mass index (BMI; kg/m2), age, sex (0 = male [referent], 1 = female), race (0 = non-Hispanic white [referent], 1 = other), and a cumulative count of comorbidities, assessed by asking participants if they had been told by a health professional that they had any of 23 different chronic conditions other than arthritis (e.g., heart disease, lung disease, back pain, osteoporosis, asthma, vision problems).

Statistical analysis.

We created adult SES and childhood SES summary scores for each participant by counting socioeconomic characteristics indicative of low SES. Not having completed high school, not owning a home, and holding a nonprofessional occupation each incremented a participant's adult SES score by 1, while having a mother who did not complete high school, having a father who held an occupation other than professional, and having parents who did not own their home during childhood each added 1 to a participant's childhood SES score. Childhood and current composite SES scores therefore both ranged from 0–3, where higher values indicated less advantageous circumstances. Participants with an SES score of 2 or 3 were collapsed into a single category due to the small number of participants with the lowest possible current SES score of 3 (n = 36 [4.6%]). As a result, current and childhood SES scores took 1 of 3 possible values: high (0 [referent]), medium ([1]), or low (2 or 3). For example, participants with a nonprofessional occupation who completed high school and who owned a home were included in the medium ([1]) current SES category (1 + 0 + 0); respondents whose mothers did not complete high school, whose fathers held a nonprofessional occupation, and whose parents were homeowners were given a low ([2]) childhood SES score (1 + 1 + 0).

All statistical analyses were performed using Stata, version 11.1. Descriptive statistics for the sample of participants were computed, stratified by current and childhood SES scores. The associations between our SES summary scores and the health outcome measures were calculated using multiple linear regression models, clustered by network practice to account for potential intrasite correlations. Unadjusted regressions of the individual SES score levels (low, medium, and high [referent]) on health outcomes were performed first, followed by 3 covariate-adjusted models: the first 2 models included levels of either current SES or childhood SES scores, and the third model simultaneously included levels of both summary scores.

Due to the complexity of the relationships between SES, sex, and race, we tested for effect measure modification of race and sex on SES summary scores in each outcome variable by adding interaction terms to adjusted linear models. Although there were no significant interactions between race and SES scores on health assessment measures, sex appeared to appreciably modify the impact of low current SES on CES-D score (P < 0.05). As a result, we stratified our models of SES on the CES-D score by sex.

RESULTS

Participants.

Socioeconomic, demographic, and health status information on our participants according to their SES summary scores is shown in Table 1. A majority of the sample participants were women (73%), had graduated from high school (87%), held a nonprofessional occupation (62%), and were homeowners (83%). There was a distinct trend toward greater SES across generations, as participants reported that their mother (or alternate caretaker) graduated from high school (46%), owned their home (70%), and held a professional occupation (20%) at significantly lower rates than they did. As shown in our stratification by SES categories, the gap between medium and high SES categories mostly reflected occupational differences, whereas the transition from medium SES to low SES involved homeownership and educational characteristics more evenly.

Table 1. Sociodemographic and health characteristics of sample participants (n = 782) within each category of current and childhood SES*
 Current SESChildhood SES
Low (n = 165)Medium (n = 358)High (n = 259)Low (n = 446)Medium (n = 231)High (n = 105)
  1. Values are the number (percentage) unless otherwise indicated. SES = socioeconomic status; BMI = body mass index; HAQ = Health Assessment Questionnaire; PCS = physical component summary (from Short Form 12 version 2 [SF-12v2]); CES-D = Center for Epidemiologic Studies Depression Scale; MCS = mental component summary (from SF-12v2).
  2. aLess than high school diploma, i.e., less than 12 years of education.
  3. bAccording to 2000 US Census codes.
Female sex136 (82)247 (69)189 (73)340 (76)157 (68)75 (71)
Not white50 (30)61 (17)21 (8)95 (21)29 (13)8 (8)
Age (range 23–94 years), mean ± SD years59.9 ± 13.959.8 ± 12.360.6 ± 13.161.9 ± 11.758.0 ± 14.257.1 ± 13.4
Education less than high schoola99 (60)6 (2)0 (0)91 (20)10 (4)4 (4)
Does not own home106 (64)28 (8)0 (0)95 (21)24 (10)15 (14)
Nonprofessional occupationb161 (98)324 (91)0 (0)329 (74)116 (50)40 (38)
Parental education less than high schoola123 (75)217 (61)84 (32)399 (89)25 (11)0 (0)
Parents did not own home80 (48)100 (28)56 (22)216 (48)20 (9)0 (0)
Parental occupation nonprofessionalb141 (85)303 (85)179 (69)437 (98)186 (81)0 (0)
SES      
High12 (7)34 (10)59 (23)92 (21)108 (47)59 (56)
Medium27 (16)96 (27)108 (42)228 (51)96 (42)34 (32)
Low126 (76)228 (64)92 (36)126 (28)27 (12)12 (11)
BMI (range 15–65), mean ± SD kg/m230.8 ± 8.2530.7 ± 6.9729.2 ± 5.8930.5 ± 7.0430.2 ± 7.0528.6 ± 6.05
Comorbidities (range 0–11), mean ± SD3.48 ± 2.203.03 ± 2.072.80 ± 1.923.40 ± 2.132.48 ± 1.792.78 ± 2.01
HAQ score (range 0–3), mean ± SD0.85 ± 0.650.69 ± 0.660.53 ± 0.560.76 ± 0.650.59 ± 0.600.46 ± 0.54
PCS score (range 0–66), mean ± SD34.8 ± 12.137.9 ± 13.141.9 ± 12.336.4 ± 12.840.6 ± 12.543.4 ± 11.9
CES-D score (range 0–55), mean ± SD15.6 ± 12.612.6 ± 12.08.06 ± 9.6113.4 ± 12.59.41 ± 9.849.21 ± 10.6
MCS score (range 0–75), mean ± SD49.6 ± 12.651.1 ± 11.153.9 ± 9.0651.0 ± 11.553.0 ± 10.052.1 ± 9.95

The mean ± SD age of the participants was 60 ± 13 years, the mean ± SD BMI was 30.2 ± 6.9 kg/m2, and they had a mean ± SD of 3.0 ± 2.1 comorbidities. Mean ± SD health scores were 0.67 ± 0.63 on the HAQ, 38.59 ± 12.92 on the PCS, and 51.76 ± 10.92 on the MCS. Women had a mean ± SD CES-D score of 12.5 ± 12.2, which was significantly higher than the mean ± SD CES-D score of 9.5 ± 9.5 for men (P < 0.01). Significant correlations existed between depressive symptoms and the other health outcomes (ranging from 0.34 with physical health to 0.72 with mental health; P < 0.01) and between physical health and disability (0.72; P < 0.01).

Regression analyses.

In unadjusted linear models (data not shown), having a low or medium current SES was associated with higher scores on the HAQ by 0.16 and 0.32 (P < 0.05), lower scores on the MCS by 2.80 and 4.33 (P < 0.01), and lower scores on the PCS by 4.04 and 7.13 (P < 0.01), respectively. Similar relationships were observed between childhood SES and measures of physical health: participants with a medium and low childhood SES scored higher on the HAQ by 0.13 and 0.30 (P < 0.01) and lower on the PCS by 2.72 and 6.94 (P < 0.01), respectively; conversely, there were no significant associations between childhood SES and MCS scores. Adjusting for covariates lowered parameter estimates, but did not appreciably affect statistical significance for the observed relationships (Tables 2 and 3). Associations of either SES summary score with health outcomes were most attenuated by covariate adjustments among participants with low SES, particularly with regard to HAQ and PCS scores. Changes were less appreciable at the medium SES level; of note, the association of medium childhood SES with PCS score became statistically significant upon covariate adjustment (P < 0.05).

Table 2. Adjusted parameter estimates and 95% CIs for the current SES summary variable associated with HAQ and Short Form 12 version 2 (PCS and MCS) scores in persons with self-reported doctor-diagnosed arthritis*
 HAQ score (n = 782)PCS score (n = 770)MCS score (n = 770)
β (95% CI)Pβ (95% CI)Pβ (95% CI)P
  1. All models adjusted for sex, age, race, body mass index, and comorbidities. Participants in the referent current socioeconomic status (SES) category had ≥12 years of education, owned a home, and had a professional occupation. 95% CI = 95% confidence interval; HAQ = Health Assessment Questionnaire; PCS = physical component summary; MCS = mental component summary.
Medium0.11 (0.01, 0.21)0.03−3.2 (−5.14, −1.26)< 0.01−2.37 (−4.13, −0.6)0.01
Low0.18 (0.04, 0.33)0.01−4.92 (−7.44, −2.41)< 0.01−3.25 (−5.5, −1.01)< 0.01
Table 3. Adjusted parameter estimates and 95% CIs for the childhood SES summary variable associated with HAQ and Short Form 12 version 2 (PCS and MCS) scores in persons with self-reported doctor-diagnosed arthritis*
 HAQ score (n = 782)PCS score (n = 770)MCS score (n = 770)
β (95% CI)Pβ (95% CI)Pβ (95% CI)P
  1. All models adjusted for sex, age, race, body mass index, and comorbidities. Participants in the referent childhood socioeconomic status (SES) category reported that their mother had ≥12 years of education, that their father had a professional occupation, and that their parents were homeowners. 95% CI = 95% confidence interval; HAQ = Health Assessment Questionnaire; PCS = physical component summary; MCS = mental component summary.
Medium0.13 (0.04, 0.22)< 0.01−2.76 (−5.34, −0.19)0.040.51 (−2.32, 3.33)0.71
Low0.15 (0.05, 0.24)< 0.01−4.32 (−6.87, −1.78)< 0.01−1.35 (−3.58, 0.89)0.22

We also examined whether SES summary scores were associated with disability, physical health, and mental health independently of each other (Table 4). Although current and childhood SES somewhat attenuated each other's effect on every outcome variable, previously observed associations of SES scores with measures of disability and physical health remained significant, with the exception of medium adult SES and PCS scores. Only low current SES remained associated with mental health.

Table 4. Concurrently adjusted parameter estimates and 95% CIs for the current and childhood SES summary variables associated with HAQ and Short Form 12 version 2 (PCS and MCS) scores in persons with self-reported doctor-diagnosed arthritis*
 HAQ score (n = 782)PCS score (n = 770)MCS score (n = 770)
β (95% CI)Pβ (95% CI)Pβ (95% CI)P
  1. All models adjusted for sex, age, race, body mass index, and comorbidities. 95% CI = 95% confidence interval; HAQ = Health Assessment Questionnaire; PCS = physical component summary; MCS = mental component summary.
  2. aParticipants in the referent current socioeconomic status (SES) category had ≥12 years of education, owned a home, and had a professional occupation.
  3. bParticipants in the referent childhood SES category reported that their mother had ≥12 years of education, that their father had a professional occupation, and that their parents were homeowners.
Current SESa      
Medium0.1 (−0.01, 0.21)0.07−2.7 (−4.74, −0.66)0.01−2.13 (−4.02, −0.24)0.03
Low0.17 (0.02, 0.33)0.03−4.36 (−6.92, −1.8)< 0.01−2.92 (−5.14, −0.69)0.01
Childhood SESb      
Medium0.12 (0.03, 0.22)0.01−2.53 (−5.24, 0.18)0.070.69 (−2.22, 3.6)0.63
Low0.1 (0, 0.21)0.04−3.17 (−5.79, −0.54)0.02−0.51 (−2.9, 1.88)0.66

In unadjusted models stratified by sex, we found that low current SES and medium current SES were associated with respective score increases of 5.5 and 8.7 on the CES-D among women (P < 0.01), whereas no significant association could be found among men. Having a low SES in childhood was likewise associated with an increase in CES-D score of 4.5 among women (P < 0.01), but not among men (data not shown). Adjusting for BMI, race, age, and comorbidities somewhat reduced parameter estimates, but did not otherwise modify associations of SES with CES-D scores (Table 5). The association of low childhood SES with greater depressive symptoms appeared to be explained by current SES in mutually adjusted models (Table 5), whereas low and medium current SES both remained associated with CES-D scores.

Table 5. Singly and concurrently adjusted parameter estimates and 95% CIs for the current and childhood SES summary variables associated with CES-D scores in persons with self-reported doctor-diagnosed arthritis, stratified by sex*
 Singly adjusted modelsMutually adjusted models
Male, β (95% CI)PFemale, β (95% CI)PMale, β (95% CI)PFemale, β (95% CI)P
  1. All models adjusted for age, race, obesity, and comorbidities. 95% CI = 95% confidence interval; CES-D = Center for Epidemiologic Studies Depression Scale.
  2. aParticipants in the referent current socioeconomic status (SES) category had ≥12 years of education, owned a home, and had a professional occupation.
  3. bParticipants in the referent childhood SES category reported that their mother had ≥12 years of education, that their father had a professional occupation, and that their parents were homeowners.
Current SESa        
Medium1.85 (−3.07, 6.77)0.444.57 (2.21, 6.94)< 0.011.3 (−3.58, 6.18)0.583.97 (1.72, 6.23)< 0.01
Low1.13 (−5.93, 8.19)0.747.52 (4.71, 10.32)< 0.010.86 (−5.88, 7.6)0.796.6 (3.94, 9.26)< 0.01
Childhood SESb        
Medium2.37 (−2.02, 6.76)0.27−0.45 (−3.77, 2.87)0.782.23 (−2.05, 6.5)0.29−0.7 (−4.1, 2.7)0.67
Low3.59 (−1.48, 8.66)0.153.65 (1.03, 6.28)< 0.013.18 (−1.33, 7.7)0.161.75 (−0.98, 4.49)0.2

DISCUSSION

Our data are consistent with associations of current and childhood SES with health outcomes in persons with arthritis, which are not fully explained by adjustments for covariates. SES summary scores attenuated each other's influences on health; however, childhood SES was independently associated with HAQ and SF-12v2 physical health scores, with current SES remaining associated with each health outcome measure.

Previous work has broadly implicated adult socioeconomic characteristics in the health of persons with arthritis ([23-25]), but childhood SES remains to be investigated as a determinant of health outcomes within that population. Our results are nevertheless in agreement with the associations of low childhood SES with poor physical health reported across population studies ([6, 7, 26, 27]). In contrast, our findings inconsistently reflect the building consensus that childhood circumstances lastingly influence mental health ([8, 9, 28]); the literature remains equivocal regarding the effects of childhood SES independently of more proximal circumstances, and our lack of positive results is in line with studies of comparable, elderly populations ([29, 30]). Differences according to sex in the effects of SES on depressive symptoms are consistent with an expanding body of research suggesting that environmental stressors to which women with low SES are disproportionately exposed, including physical abuse and single parenthood, may put them at a higher risk of depression than men with low SES ([31, 32]).

Since the HAQ, CES-D, and SF-12 are generic health assessment tools, our results may be explained by mechanisms not specific to arthritis in addition to ones more closely associated with the disease. Specifically, the observed associations of SES with health outcomes were likely mediated by socioeconomic patterns of health behaviors such as alcohol use, diet consumption, physical activity, and smoking habits ([33-37]); in particular, physical activity is a strong determinant of physical health outcomes independently of BMI among persons with arthritis, as shown by the consensus on the value of physical exercise in the management of arthritis ([38]). Socioeconomic differences in environmental exposures and in the utilization of health care may also explain our results; with regard to physical health and arthritis, individuals with low SES are more likely to sustain musculoskeletal injuries ([39]), and the use of procedures such as joint replacement surgeries follows a positive gradient ([40]). Health disparities are furthermore known to have a psychosocial dimension, and SES-associated characteristics such as self-efficacy may influence the health outcomes of chronic pain disorders such as arthritis ([41]). Biologically, exaggerated inflammatory reactions among individuals with low SES have been documented ([42, 43]), and adverse life events associated with low SES have been found to dysregulate hypothalamic–pituitary–adrenal axis activity, which is closely tied into chronic pain ([44]).

Similar mechanisms may explain independent associations of low childhood SES with physical health outcomes insofar as poor childhood health, for instance, due to inadequate medical care or to harmful environmental exposures, may have a lasting impact on the life course ([17]). Studies have also stressed the importance of behaviors taking their roots in early life stages as a pathway by which childhood socioeconomic characteristics may affect future health ([45, 46]). Adverse circumstances may furthermore have an amplified effect during critical developmental phases such as childhood, as underscored by the increasingly well-documented association of child abuse with the development of OA in adulthood ([47]) and by epigenetics research suggesting that adverse early socioeconomic circumstances may trigger harmful defensive phenotypes associated with chronic health ([48]).

To the best of our knowledge, this investigation is the first to specifically examine associations of childhood SES with health outcomes among persons with arthritis. The validity of our findings is supported by our use of well-established measures of health outcomes and by our adjusting for potential confounder variables. Conversely, methodologic limitations include the inherent imprecision involved in measuring SES as well as possible recall bias in our retrospective assessment of childhood SES; however, a standard approach to comprehensively summarize SES does not currently exist, and recollections of parental socioeconomic characteristics are generally considered to be reliable ([49]). Additionally, this study's cross-sectional design precludes us from inferring causal relationships, particularly with regard to current SES and health outcomes. It should also be noted that although OA likely affected most individuals in this primary care sample, grouping participants into a single disease category certainly overlooked meaningful differences between the many types of arthritis. Furthermore, recruiting participants from family practices excluded individuals who did not visit a primary care provider, perhaps for reasons tied to their low SES. Although our self-reported outcome measures are considered to adequately reflect health status, it has been suggested that self-reported outcomes such as SF-12v2 items may underestimate health disparities due to reporting heterogeneity according to SES, although the direction and magnitude of the effect remain a matter of debate ([50]).

This investigation underscores the role that SES plays in the health status of persons with arthritis over the entire life course. Studies on health disparities should evaluate socioeconomic circumstances across life stages, including childhood. It may be beneficial for primary care providers to inquire about the past and present SES of their patients with arthritis, insofar as such characteristics may inform the assessment of risk factors and health outcomes.

It would be worthwhile to examine whether the associations we observed likewise occur in other, potentially larger, samples of individuals with arthritis, and to determine if the health and SES connection varies across specific arthritis disorders. Investigating the extent to which behaviors and adverse life events explain these health disparities appears to be a logical next step in this line of research. Our study adds to the consensus that intervening early in the life course may pay health dividends into old age; however, since greater insight into causal mechanisms is lacking, it remains unclear what policies may best mitigate the adverse impact of low childhood SES among persons with arthritis.

AUTHOR CONTRIBUTIONS

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. Callahan 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. Baldassari, Cleveland, Callahan.

Acquisition of data. Callahan.

Analysis and interpretation of data. Baldassari, Cleveland, Callahan.

ACKNOWLEDGMENTS

The authors would like to thank Robert DeVellis, Jay Kaufman, Thelma Mielenz, Randy Randolph, Britta Schoster, Phillip Sloane, Todd Schwartz, and Morris Weinberger for their contributions and input to the study. We thank the following participating family practices in the NC-FM-RN for their assistance: Black River Health Services, Burgaw; Bladen Medical Associates, Elizabethtown; Blair Family Medicine, Wallace; Cabarrus Family Medicine, Concord; Cabarrus Family Medicine, Harrisburg; Cabarrus Family Medicine, Kannapolis; Cabarrus Family Medicine, Mt. Pleasant; Chatham Primary Care, Siler City; CMC Biddle Point, Charlotte; CMC North Park, Charlotte; Community Family Practice, Asheville; Cornerstone Medical Center, Burlington; Dayspring Family Medicine, Eden; Family Practice of Summerfield, Summerfield; Goldsboro Family Physicians, Goldsboro; Henderson Family Health Center, Henderson; Orange Family Medical Group, Hillsborough; Person Family Medical Center, Roxboro; Pittsboro Family Medicine, Pittsboro; Prospect Hill Community Health Center, Prospect Hill; Robbins Family Practice, Robbins; and Village Family Medicine, Chapel Hill. Finally, we thank the individuals who willingly participated in the study.

Ancillary