Impact of socioeconomic status on the course of rheumatoid arthritis and on related use of health care services
To quantify the impact of socioeconomic status (SES) among patients with rheumatoid arthritis on 1) health outcomes and related health care utilization in relation to disease duration and 2) changes in health outcomes and related health care utilization over a 2-year period.
A questionnaire survey was conducted among 878 patients with rheumatoid arthritis (RA), varying in disease duration from 0 to more than 15 years. To determine the impact of SES on the health outcomes and health care use, patients were compared within and between 3 disease duration groups. Additionally, longitudinal changes in health outcomes and health care use were assessed with a followup questionnaire sent out 2 years later.
Patients with low SES have worse disease activity, physical health, mental health, and quality of life than patients with high SES. These differences, however, decreased over time. Regarding health care use, we found that patients with low SES made considerably less use of allied health care than patients with high SES.
Efforts should be undertaken in health care to alleviate the health disadvantages of RA patients in lower socioeconomic groups. In particular, the access to allied health care could be improved.
Rheumatoid arthritis (RA) is a chronic and disabling disease that has a major impact on the lives of patients. Many patients have impaired physical health and are dependent on a wide variety of health care services (1).
Among individuals with a low socioeconomic status (SES), RA is more prevalent (2), the impact of the disease is higher (3–9), and the risk of mortality is higher (3, 10, 11). Because of these disadvantages, it is said that patients with a low SES are exposed to double suffering, having both a higher prevalence of RA and more severe RA (12).
Generally, we would expect that patients with low SES, considering they are less healthy, would use more health care. When equal access to health care is considered, the variations in the use of health care between low and high SES groups should disappear when health-related needs are taken into account (13). Equity in health care assumes that utilization of care is predominantly determined by health characteristics, because health care should be provided irrespectively of SES; inequity in health care is present if the use of care is related to enabling factors, such as SES.
With regard to the use of health care among patients with RA, it was found that socioeconomic deprivation did not lead to systematic differences in medication treatment (8, 14, 15). Moreover, it is not clear whether socioeconomic differences are present in the use of health care services among patients with RA, and whether differences in health outcomes and in use of health care decrease or increase during the course of the disease. On the one hand, differences in SES in long-term health outcomes and long-term health care utilization may decrease because later in life the survivors with a low SES represent a relatively healthy subgroup (16). On the other hand, socioeconomic differences in health outcomes and health care use may increase because of cumulative advantages experienced during life span (17). Provided that differences in SES are present, knowledge on this subject is important to determine in which part of the care process problems exist, arise, or expand. These results could be used to adjust the care process to assure equal use of care for equal needs, independent of SES.
The aim of this article is to quantify the impact of SES among patients with RA on 1) health outcomes and health care utilization in relation to disease duration and 2) changes in health outcomes and changes in health care utilization over a 2-year period.
PATIENTS AND METHODS
A sample of 1,200 patients with RA, stratified by disease duration, was selected from the register of a large rheumatology outpatient clinic in Amsterdam, which included the patients from 5 allied outpatient clinics. The patients were followed prospectively from 1997 by means of 2 postal questionnaires with a 2-year interval, combined with a clinical examination. Inclusion criteria were at least 16 years of age and fulfilling the 1987 revised American College of Rheumatology (formerly American Rheumatism Association) criteria for RA (18). All patients gave informed consent. The medical ethical committee approved the study design.
Of the 1,200 eligible patients, 878 patients (73%) took part in the study at baseline. No significant differences were found between respondents and nonrespondents with respect to sex and education level (19), although nonrespondents were older. Nine patients were excluded from analyses because no information was available on SES. Of the remaining 869 respondents, 674 patients (76%) completed the 2-year followup. Between these respondents (n = 674) and the nonrespondents (n = 195), no differences were found regarding SES, age, or sex. Patients who were lost to followup because of death or relocation were significantly older (P = 0.001) than respondents, but no differences were found regarding SES or sex.
From the questionnaires, clinical examinations, and patient records, information was extracted on sociodemographic variables, health characteristics, and health care utilization. Sociodemographic variables were age, sex, and socioeconomic status. SES was assessed by formal education level, and divided into 3 groups: low SES, including patients with no education or primary school level; medium SES, including patients with secondary school level; and high SES, including patients with college or university level.
Health characteristics included were disease duration, disease activity, physical health, mental health, and quality of life. Patients were subdivided into 3 disease duration groups (0–5 years, 5–15 years, and ≥15 years). Disease activity was calculated according to the Disease Activity Score in 28 swollen and 28 tender joints (DAS28), without the visual analog scale (VAS) on general health (20). Three disease activity subgroups were formed. The cutoff points for the various disease activity groups were recalculated from the cutoff points of the original Disease Activity Score (21) into cutoff points for the DAS28 (22) resulting in the following 3 groups: low disease activity (DAS28 ≤ 3.3), moderate disease activity (DAS28 > 3.3 and DAS28 ≤ 5.1), and high disease activity (DAS28 > 5.1). For analysis, the groups of moderate and high disease activity were combined (DAS28 > 3.3). Physical health was assessed according to the Dutch version of the Health Assessment Questionnaire (HAQ) (23), and mental health according to the Dutch version of the Center for Epidemiological Studies Depression Scale (CES-D) (24). Physical health was classified into 2 disability groups: none or minor disabilities (HAQ = 0–0.5) and major disabilities (HAQ = 0.5–3.0). The threshold for possible cases of depression on the mental health measure given by the CES-D manual was 16. Thus, 2 groups were formed: no depression (CES-D < 16) and possible depression (CES-D ≥ 16). Quality of life was assessed by means of a 100-mm VAS ranging from 0, which equals the worst possible quality of life, to 100, which equals the best possible quality of life (25). The patients were divided into 2 groups by means of the baseline median (i.e., 62 mm).
Health care utilization in relation to RA was assessed with respect to the preceding 12 months for a wide variety of health care providers. The care providers were aggregated as follows: rheumatology care (i.e., rheumatologist or general practitioner), additional specialist care (i.e., rehabilitation specialist, orthopedic surgeon, or plastic surgeon), allied health care (i.e., physiotherapist, occupational therapist, or chiropodist), psychosocial care (i.e., social worker, psychologist, or psychiatrist), and home nurse care.
To evaluate the compilation of SES in the patient population, we compared the SES of patients in the study population with the SES of the general Dutch population, according to education level, adjusted for age and sex. By means of chi-square statistics, differences between the expected and the observed count of patients in all SES groups were assessed. Results were considered statistically significant when P values were below 0.05.
The impact of SES on health outcomes of patients was assessed within 3 disease duration groups (0–5 years, 5–15 years, and ≥15 years) by means of multivariable logistic regression analyses. The continuous variables (i.e., disease activity, physical health, and quality of life) were first assessed by linear regression analysis, and thereafter by logistic regression, dichotomized as described before. The linear and the logistic models yielded similar results. For each disease duration group separately, a multivariable logistic model was built, containing SES, and age, sex, and disease duration for adjustment. In addition, a multivariable model was constructed to determine the impact of SES on the course of health outcomes over disease duration. This model contained SES and disease duration and their interaction (trend variable) as independent variables and each of the health outcomes as the dependent variable, including age and sex, for adjustment.
The impact of SES on health care use was assessed similarly as described for health outcomes. The logistic models with health care utilization as the dependent variable were adjusted not only for age, sex, and disease duration, but also for all health outcomes (i.e., disease activity, physical health, mental health, and quality of life).
The effect sizes from the logistic models were expressed as odds ratios (ORs) with 95% confidence intervals (95% CIs). Analyses were performed using SPSS 10.0.7 for Windows (SPSS, Chicago IL).
Longitudinal changes in health outcomes and in health care utilization were calculated over a 2-year period, independent of disease duration.
Changes in health outcomes were determined as follows. Clinically relevant changes in disease activity were defined as a decrease or increase of the baseline score of 1.2 or more (26). Clinically relevant changes in physical health were defined as a decrease or increase of the baseline score of 0.48 or more (27). Because evidence-based meaningful changes on the quality of life VAS are unknown, we defined a meaningful change in quality of life as a decrease or increase of the baseline score of 1 standard deviation (i.e., 22 mm) or more to cover a high probability of the true change on the quality of life measure (28). Clinical changes in mental health were determined when patients changed from no depression to depressive symptoms or vice versa. Patients not able to improve or worsen, because they already were in the best or worst possible state, were excluded from the respective analyses.
Changes in the use of health care were calculated when a patient changed from no use to use of an aggregated group of health care providers or vice versa. Patients not able to decrease or increase the use of health care, because they were not using or already using care, were excluded from the respective analyses.
The effect sizes of the logistic models were expressed as ORs with 95% CIs. Analyses were performed using SPSS 10.0.7 for Windows.
The characteristics of the respondents at baseline (n = 869), divided by SES and disease duration, are shown in Table 1. The low SES group had 220 patients (25%); 526 patients (61%) were in the medium SES group, and 123 patients (14%) were in the high SES group. The mean disease duration of the patients was 8.7 years (SD 9.2). The mean age was 59.5 years (SD 13.8), and 71% were women. Patients with a low SES were older than patients with a high SES (P < 0.001). Disease duration was found to be significantly different between SES groups (P < 0.001). A relatively high disease duration was found among patients in the medium SES group (9.0 years) compared with patients in the low and high SES groups (8.4 years and 8.2 years, respectively).
Table 1. Characteristics of patients with rheumatoid arthritis (n = 869) at baseline by socioeconomic status and disease duration for cross-sectional analyses*
|Age, years||67.8 (12.4)||55.9 (15.0)||49.0 (14.1)||66.6 (12.4)||57.6 (13.6)||53.5 (12.5)||71.2 (11.4)||62.8 (13.2)||59.6 (13.2)|
|Female||80 (74)||182 (73)||37 (61)||38 (63)||122 (73)||29 (58)||43 (83)||84 (76)||10 (83)|
|High disease activity (DAS28)||56 (67)||107 (51)||19 (38)||21 (54)||81 (59)||24 (56)||33 (89)||52 (70)||6 (75)|
|Major disabilities (HAQ)||50 (47)||90 (36)||16 (26)||36 (60)||79 (47)||20 (40)||41 (80)||75 (68)||9 (75)|
|Possible depression (CES-D)||32 (31)||67 (27)||11 (18)||21 (36)||47 (28)||11 (22)||24 (49)||31 (29)||4 (33)|
|Low quality of life (VAS)||63 (58)||138 (55)||21 (34)||37 (62)||84 (50)||16 (32)||33 (63)||47 (43)||3 (25)|
| Rheumatology care||98 (91)||243 (98)||59 (97)||58 (97)||164 (98)||48 (96)||50 (96)||108 (98)||12 (100)|
| Additional specialist care||55 (51)||121 (49)||26 (43)||28 (47)||102 (61)||29 (58)||29 (56)||72 (66)||10 (83)|
| Allied health care||44 (41)||118 (47)||36 (59)||23 (38)||89 (53)||28 (56)||25 (48)||57 (52)||7 (58)|
| Psychosocial care||11 (10)||35 (14)||11 (18)||5 (8)||22 (13)||6 (12)||4 (8)||13 (12)||2 (17)|
| Nurse care||2 (2)||2 (1)||0 (0)||4 (7)||11 (7)||3 (6)||3 (6)||9 (8)||2 (17)|
The SES distribution in the patient population, adjusted for age and sex, differed significantly from the general Dutch population (29). The number of observed patients (NOb) in the low SES group was significantly less than the expected number (NEx) of patients (NOb = 220 versus NEx = 291; χ2 = 17.5, P < 0.001), whereas the medium SES group contained more patients than expected (NOb = 526 versus NEx = 456; χ2 = 10.8, P < 0.005). For the high SES group, no significant difference was found between the observed and the expected number of patients (NOb = 123 versus NEx = 122).
The health outcomes of patients varied, in each disease duration group, between SES groups (Table 1). We found that in the group with a disease duration of up to 5 years, patients with low SES presented significantly worse health outcomes than patients with high SES (Table 2). Compared with patients with high SES, those in the low SES group suffered more often from a high disease activity (OR = 2.8, 95% CI 1.2–6.4), more often had disabilities (OR = 2.2, 95% CI 1.0–4.6), reported depressive symptoms more often (OR = 2.6, 95% CI 1.1–6.0), and perceived a low quality of life more often (OR = 3.1, 95% CI 1.5–6.4).
Table 2. Impact of SES, according to the cross-sectional analyses at baseline, on the course of health and use of health care services of patients with rheumatoid arthritis (n = 869)*
|High disease activity (DAS28 >3.3)||Low||3.3 (1.6–6.7)||1.3 (0.5–3.0)||1.5 (0.2–10.3)|
| ||Medium||1.5 (0.8–2.6)||1.0 (0.5–2.1)||0.7 (0.1–3.5)|
|Major disabilities (HAQ ≥0.5)||Low||2.2 (1.0–4.6)||1.8 (0.8–4.2)||1.0 (0.2–4.7)|
| ||Medium||1.5 (0.8–2.8)||1.1 (0.6–2.2)||0.7 (0.2–2.8)|
|Possible depression (CES-D ≥16)||Low||2.6 (1.1–6.0)||2.0 (0.8–5.0)||2.1 (0.5–8.2)|
| ||Medium||1.8 (0.9–3.7)||1.2 (0.6–2.7)||0.9 (0.2–3.2)|
|Low quality of life (VAS <62)||Low||3.1 (1.5–6.4)||2.2 (1.0–5.0)||3.8 (1.0–15.0)|
| ||Medium||2.2 (1.2–4.0)||1.6 (0.8–3.0)||1.8 (0.5–6.4)|
|Use of Rheumatology care†||Low||0.3 (0.0–4.5)||—||—|
| ||Medium||2.1 (0.2–26.2)|| || |
| ||High||1.0|| || |
| Additional specialist care‡||Low||1.0 (0.4–2.3)||0.7 (0.2–1.8)||0.3 (0.0–3.2)|
| ||Medium||1.2 (0.6–2.4)||1.1 (0.5–2.3)||0.3 (0.0–2.9)|
| Allied health care||Low||0.3 (0.1–0.8)||0.5 (0.2–1.5)||0.8 (0.1–4.8)|
| ||Medium||0.5 (0.2–1.0)||0.8 (0.4–1.6)||0.8 (0.2–3.9)|
|Psychosocial care||Low||0.4 (0.1–1.5)||1.5 (0.2–9.2)||3.5 (0.1–139.0)|
| ||Medium||0.5 (0.2–1.2)||1.1 (0.3–4.1)||0.6 (0.0–18.0)|
|Home nurse care†||Low||—||0.2 (0.0–2.1)||0.1 (0.0–3.5)|
| ||Medium|| ||0.5 (0.1–2.3)||0.2 (0.0–4.5)|
| ||High|| ||1.0||1.0|
Within the group with a 5–15-year disease duration, no statistically significant differences in health outcomes were observed between patients with a low SES and a high SES. Among patients with a disease duration of 15 years or more, also no statistically significant impact of SES on health outcomes was found. Although some trends in health outcomes seem to exist over disease course (Table 2), the trend analysis showed that no statistically significant differences in health outcomes were present between disease duration groups.
Regarding the use of health care, we observed that patients with a low SES and 0–5-years disease duration received significantly less allied health care than patients with a high SES with comparable disease duration. For the use of other health care services, no statistically significant impact of SES was found. The results of the trend analysis, however, showed that patients with a low SES made significantly less use of additional specialist care with increasing disease duration in comparison with patients with a high SES. For the use of other health care services, no significant trends were observed.
At the 2-year followup in each of the SES groups, more patients reported disabilities and depressive symptoms than at baseline (Table 3). Regarding disease activity and quality of life, patients with low SES improved in the 2-year period. Patients with high SES, however, improved on disease activity, but worsened on quality of life.
Table 3. Characteristics of patients with rheumatoid arthritis (n = 674) at baseline and after 2 years, by socioeconomic status for longitudinal analyses*
|Age, years||67.3 (11.6)||69.3 (11.6)||58.2 (13.5)||60.2 (13.5)||52.3 (13.1)||54.3 (13.1)|
|Female||117 (72)||117 (72)||303 (74)||303 (74)||62 (62)||62 (62)|
|Disease duration, years||8.4 (86)||10.4 (86)||9.0 (9.8)||11.0 (9.8)||8.2 (8.3)||10.2 (8.3)|
|High disease activity (DAS28)||78 (66)||65 (62)||196 (56)||184 (59)||43 (49)||28 (37)|
|Major disabilities (HAQ)||86 (53)||94 (59)||179 (43)||208 (50)||37 (37)||46 (46)|
|Possible depression (CES-D)||51 (33)||53 (35)||101 (25)||117 (29)||17 (17)||17 (17)|
|Low quality of life (VAS)||101 (62)||92 (60)||206 (50)||193 (48)||26 (26)||29 (29)|
|Rheumatology care use||154 (95)||150 (93)||404 (98)||396 (97)||96 (96)||94 (95)|
|Additional specialist care use||85 (52)||52 (32)||229 (56)||134 (33)||55 (55)||38 (38)|
|Allied health care use||73 (45)||70 (43)||212 (51)||174 (42)||57 (57)||51 (52)|
|Psychosocial care use||15 (9)||9 (6)||49 (12)||32 (8)||15 (15)||10 (10)|
|Nurse care use||5 (3)||6 (4)||15 (4)||18 (4)||5 (5)||6 (6)|
Multivariable logistic analyses on the 2-year followup showed, with respect to worsening in health outcomes, no statistically significant differences between SES groups. With respect to improvements, we found that significantly more patients with a low SES improved over a 2-year period in quality of life (OR = 2.7, 95% CI 1.2–6.3) than patients with a high SES (Table 4).
Table 4. Impact of SES on worsening and improvement in health among patients with rheumatoid arthritis (n = 674) during a 2-year followup*
|Worsened in 2 years|| || || || |
| Low SES||1.9 (0.7–5.3)||1.4 (0.6–3.2)||1.0 (0.4–2.5)||1.0 (0.4–2.3)|
| Medium SES||1.3 (0.6–2.9)||1.1 (0.5–2.2)||1.3 (0.6–2.6)||1.2 (0.6–2.5)|
| High SES||1.0||1.0||1.0||1.0|
|Improved in 2 years|| || || || |
| Low SES||1.4 (0.6–3.6)||1.1 (0.3–4.3)||0.4 (0.1–1.3)||2.7 (1.2–6.3)|
| Medium SES||1.0 (0.5–2.0)||1.1 (0.4–3.5)||0.5 (0.2–1.2)||1.8 (0.9–3.7)|
| High SES||1.0||1.0||1.0||1.0|
No indications were found for a relationship between SES and changes in health care use.
In this study, we quantified the impact of SES among patients with RA on 1) health outcomes and health care utilization in relation to disease duration and 2) changes in health outcomes and changes in health care utilization over a 2-year period.
We analyzed health outcomes and health care utilization by means of cross-sectional and longitudinal analyses. Although cross-sectional analyses may be potentially more biased than longitudinal analyses for detecting changes in outcome measures, they have the advantage of creating a wider time span in the data.
We showed that the distribution of SES in this patient group, adjusted for age and sex, differed significantly from the general Dutch population (29). Taking into account that RA has a higher prevalence among individuals with low SES (2), one would expect more patients in the low SES group compared with the general population. In contrast, the low SES patient group contained considerably fewer patients than was expected. This might indicate a problem in the access to care of patients with low SES, either in the access to primary care or in the referral from primary care to secondary or tertiary care centers.
The results of this study confirmed findings of previous studies (3, 5, 6): patients with a low SES have worse health outcomes than patients with a high SES. However, significant differences in health outcomes between patients with high SES and patients with low SES are only present in the 0–5-year disease duration group. Irrespective of SES, patients tend to fail in health over time, in an absolute way, on the number of disabilities and on the prevalence of depressive symptoms. Relative differences in these health outcomes, however, decline between SES groups over time, as patients with a low SES grow worse less quickly than patients with a high SES.
Over a 2-year period, patients with low SES worsened on the more objective measures, i.e., physical health (HAQ) and mental health (CES-D), but improved more often than those with high SES on the overall measure of quality of life. These results indicate that patients with low SES might have a different perception or interpretation of quality of life than patients with high SES.
The principle of equity in health care presupposes that the use of health care is influenced by health characteristics and not by SES. Equity in health care was demonstrated by this study for most health care services. We found, however, that patients with low SES used less allied health care than patients with high SES, after adjusting for age, sex, and health outcomes. In the Netherlands, no financial barriers in health care are present because of regulations of the Dutch health insurance system. Therefore, no financial access problems to allied health care could be expected. Our results suggest, as mentioned in previous research (8), that higher educated people might be better equipped to negotiate for medical care, while they are not necessarily more in need of health care.
We think that the strength of this study is the longitudinal design with a large group of patients with a wide span in disease duration, combined with comprehensive assessments of health outcomes and health care utilization. However, some remarks can be made. First, a followup period of 2 years is short for detecting a SES gradient on changes in health outcomes and in the use of health care services over time, especially in the case of a progressive chronic condition like RA. Second, we might have overestimated the SES differences in health outcomes in our study, especially among women, because we used formal education level of the patient to assign patients to the various SES groups. We did not take into account that patients could benefit from the socioeconomic circumstances of a higher educated partner. Finally, a bias might be present in our data because no information is available on patients with RA who have not (yet) been seeking care. A bias occurs if patients from lower socioeconomic groups with milder onset RA are less likely to seek medical advice, in comparison with patients from privileged groups, resulting in an overestimation of SES differences.
In conclusion, patients with low SES have worse health outcomes than patients with high SES, but during the course of the disease the differences in health outcomes between SES groups decrease. Regarding most health care services, equity in use was demonstrated. With respect to the use of allied health care, however, we found that patients with a low SES used considerably less care than patients with a high SES. Thus, efforts should be undertaken in health care to alleviate the health disadvantages of RA patients in lower socioeconomic groups. In particular, the access to allied health care could be improved for RA patients with low SES.
We thank Marga Kammeijer for her large contribution to the collection of data.