METHODS

Study design and data collection

This cross sectional, Institutional Review Board approved study assessed English or Chinese speaking Chinese, Malay, or Indian inpatients or outpatients with knee or hip OA attending consecutively over two weeks at Tan Tock Seng Hospital, a tertiary referral hospital in Singapore, an Asian country with an urban population of 3.3 million Chinese, Malays, and Indians (ethnic distribution: 76.8, 13.9, and 7.9% respectively). Inclusion criteria were (a) knee or hip OA; (b) absence of other forms of arthritis or lower limb conditions causing immobility. Patients with hip or knee OA were studied together as these conditions tend to coexist, and so that our results would be comparable with other studies.^3,8 Trained fieldworkers conducted a structured interview and medical record review using a pre-tested data collection form. Subjects answered questions with reference to the most symptomatic knee or hip joint (that is, the index joint). Two trained examiners unaware of the subject data examined patients to grade the severity of OA using a visual analogue scale and to assess the American College of Rheumatology functional status.⁹

Instruments and definitions

In the Short-Form 36 (SF-36) bodily pain (BP) and physical functioning (PF) scales,^10,11 higher scores (range 0–100) reflect less pain or better physical functioning, respectively. Higher scores in the Helplessness Subscale indicate more learned helplessness, in which, as a result of adverse past experiences, subjects believe their efforts will be ineffective.¹² Higher scores in the Family Functioning Measure¹³ reflect better interactions among family members. OA was diagnosed by a rheumatologist or orthopaedic surgeon using clinical and radiographic assessment. Comorbidities were self reported diabetes, hypertension, stroke, or ischaemic heart disease. Ethnicity was that stated in a subject’s national registration identification card.

Statistical analyses

The association of BP or PF scores with possible factors affecting pain or physical function was assessed using Pearson’s correlations (table 1) or one way parametric analysis of variance (table 2). Exploratory analysis including all these possible factors was performed (with factors in table 2 coded using dummy variables¹⁴) to identify the best set of predictors of pain or physical function using separate linear regression models with BP and PF scores as outcome variables; a combination of forward stepwise and all possible regressionmodelling procedures was used, which yields better fitting models than stepwise regression alone.¹⁴ Questionnaire language may influence PF or BP scores and was assessed by including a language variable (English/Chinese) in the model building process. First order interactions were studied for factors identified in the model building process, and diagnostic tests were performed to confirm that assumptions for linear regression were fulfilled.

RESULTS

Subject characteristics and SF-36 scores

Most of the subjects were elderly Chinese women, and the commonest site of OA was at the knee (tables 1 and 2). English and Chinese questionnaires were completed by 59 and 67 subjects, respectively. The subjects’ mean BP and PF scores of 57.7 and 56.2, respectively, were substantially lower than the 79.3 and 80.8 points, respectively, expected in the general population (adjusted for age, sex, ethnicity, and questionnaire language).¹¹

Factors influencing pain in Asian patients with OA

All factors significantly associated with BP on univariate analysis (tables 1 and 2) were selected in the modelling process (table 3). Of note, scoring for the BP scale is such that higher scores indicate less bodily pain. Subjects who were younger, working, Chinese, with a shorter duration of limitation in normal activities or with ≥1 year of education had less pain. First order interactions were such that in Chinese subjects with ≥1 year of education, increasing age was associated with less pain, whereas in Chinese with no education, increasing age was associated with more pain. In non-Chinese subjects, increasing age was associated with increased pain for all subjects, and was more marked in subjects with no education.

Factors influencing physical function in Asian patients with OA

All factors significantly associated with PF on univariate analysis with the exception of pain on movement (tables 1 and 2) were selected in the modelling process (table 4). Subjects with less bodily pain, less severe OA, less learned helplessness, or with one or more years of education had better physical function. First order interactions between BP and education were such that in subjects with no years of education, more pain resulted in a greater reduction in physical function than in subjects with ≥1 year of education.

DISCUSSION

This study found that socioeconomic status and psychosocial factors, as well as ethnicity, demographic, and disease related factors, independently influenced pain or physical function in Asian patients in Singapore with OA. These findings have several implications. Firstly, several factors were associated with differences in PF or BP scores that exceeded five points (suggested as the minimum clinically important difference for SF-36 scores¹⁰), and may therefore have a clinically important impact on the health status of patients with OA. As some of these factors (for example, learned helplessness) may be modified, interventions in these areas may reduce pain and improve physical function in such patients. Secondly, our data show that socioeconomic status and psychosocial factors influence pain and functional status in a multiethnic, urban Asian sociocultural context. To the best of our knowledge, this has not been previously demonstrated. Interestingly, the influence of socioeconomic status on physical function was modulated by the degree of pain, highlighting the close relationship among these three domains. Thirdly, our data further highlight the fact that patients with OA in an Asian population, like those in non-Asian populations, have substantially more pain and a lower level of physical functioning than the general population.

Factors influencing pain or physical function in Asians with OA in this study were broadly similar to those identified in studies in non-Asian populations.^4–6 Modelling restricted to female Chinese subjects (performed because of the small number of male and non-Chinese subjects) yielded similar results to models using the entire study sample. Some previously reported associations (for example, between BP or PF and sex,⁴ body mass index,^4,6 or marital status⁴) were not seen in this study. There are several possible reasons for this. Firstly, specific factors influencing PF or BP may differ among various ethnic groups—for example, black and white Americans with OA⁷. Secondly, the different instruments used to assess pain or physical function in various studies might have influenced the selection of factors associated with these outcomes.⁶

Ethnicity influenced the degree of pain independently of socioeconomic status in this study (with Chinese showing clinically important differences in BP scores¹⁰), and was modulated by age and educational status, as shown by the interactions involving these variables. Ethnic differences in factors affecting physical function have previously been reported in patients with OA.⁷ Observed ethnic differences in BP scores may be related to ethnic differences in unmeasured determinants of pain arising from ethnic differences in culture, identity or minority status¹⁵ and including genetic, psychosocial, or disease related factors, as seen in patients with cancer.¹⁵

Study limitations include the cross sectional design, which, as in similar studies,^1,3–6 demonstrates association rather than causation. Studying subjects attending a tertiary care hospital and the small numbers of male and non-Chinese subjects limit the generalisability of our results (especially for ethnic differences in pain); however, our data are important in justifying a larger, population based study. Radiographic severity of OA (associated with pain or physical function in some,³ but not other studies⁶) was not assessed because recent radiographs were not available for some patients, and clinical assessment of OA severity was therefore used as a pragmatic compromise (though the validity and reliability of this approach needs further study).

In conclusion, we found that socioeconomic status and psychosocial factors independently influenced pain or physical function in Asian patients with OA in Singapore, further extending the sociocultural contexts in which this paradigm has been demonstrated. Future studies are needed to determine if this paradigm exists in other Asian sociocultural contexts, and to identify interventions to reduce pain and improve function in this common and disabling condition.