While some women with confirmed arthritis did not have recent musculoskeletal signs or symptoms, others with the signs and symptoms did not report diagnosed arthritis.

Arthritis has become the most common and principal reason of pain and disability, especially among the elderly population around the world. Government has to dedicate a considerable amount of healthcare resources for the arthritis management. The occurrence of arthritis is continuously increasing and has affected nearly 50 million people in the USA. Although it may occur in anyone, women are at greater risk. One of the most common form of arthritis is osteoarthritis, which is known to affect women more adversely and at multiple sites when compared men of their age. Thus, the rate of joint replacements is also much higher in women than men. The management of arthritis has become a real challenge for the healthcare systems with the limited resources. The major source of information for epidemiological studies and other health research is self-reported heath survey data. The reported health data is feasible because the data is routinely collected by government and/or agencies and is easily available to the common audience. Although it has been argued that self-reported diagnosis of chronic conditions may suffer from recall-bias, leading to underreporting of conditions and underestimation of incidence, some investigators have confirmed the use of self-reported arthritis as it has good agreement with medical records, and an adequate level of sensitivity and specificity in previous validation studies Previous validation studies of self-reported arthritis have mostly been based on a non-gender specific sample. Since women are most at risk of arthritis, a study with a focus on women represents an important milestone to the better understanding of the validity of self-reported arthritis and its application in large epidemiological studies.

Rationale behind the research:

• The validity of self-reported arthritis and the determinants of its accuracy among women have not been thoroughly studied
• This study was conducted to examine the accuracy of self-reported arthritis as the case-definition in community-living women for the epidemiological study of arthritis

Objective:

The objectives of this study were to:

1) Examine the agreement between self-report diagnosed arthritis and musculoskeletal signs and symptoms in community-living older women;

2) Estimate the sensitivity, specificity, and predictive values of self-reported arthritis; and

3) Assess the factors associated with the disagreement.

Study outcomes

Self-reported diagnosed arthritis: In the questionnaire based survey, participants were asked: “In the past 3 years, have you been diagnosed or treated for (a list of conditions)?” The list of condiitions (forms of arthritis) included osteoarthritis, rheumatoid arthritis, psoriatic arthritis, gout, and/or other form of arthritis. Self-reported diagnosed arthritis in the present study was defined as an answer“Yes” to any form of these forms of arthritis. Agreement between self-reported arthritis and musculoskeletal signs symptoms was measured by Cohen’s kappa. Sensitivity, specificity, and predictive values of self-reported arthritis were estimated using musculoskeletal signs and symptoms as the reference standard. Factors associated with disagreement between self-reported arthritis and the reference standard were examined using multiple logistic regression.

Outcomes:

• Overall, women with and without self-reported arthritis are not significantly different in socio-demographic characteristics; but, most of them were obese, had more chronic conditions, worse quality of life , greater disability, and worse in The Western Ontario and McMaster Universities Arthritis Index (WOMAC) total score compared to women without arthritis
• Crude prevalence of arthritis estimates based on self-reported arthritis and musculoskeletal signs and symptoms were not significantly different ((39.1% for self-reported arthritis and 34.7% for musculoskeleton signs and symptoms). Agreement between self-reported arthritis and musculoskeletal symptoms was moderate (κ=0.41, 95% CI; 0.33–0.49). Sensitivity of self-reported arthritis was 66.7% (95% CI 60.0–73.3%), whereas, specificity was 75.5% (95% CI; 71.1–79.9%).
• When using musculoskeletal signs and symptoms as the reference standard, univariate analysis results indicated that false-positiveness of self-reported arthritis was associated with a higher level of education, being obese, or better physical health; and negatively associated with disability or a higher WOMAC score. They also indicated that false-negatives were associated obesity, disability or the WOMAC score; but negatively associated with physical health. Details of the univariate analysis results can be found in an additional file.
• Results of the multiple logistic regression showed  that after controlling the socio-demographics, lifestyle variables including obesity were not significant predictors of false-positive. When controlling for both socio-demographic and lifestyle variables, better physical health, lower disability measure or WOMAC score is significantly associated with false-positive. However, when the health variables were simultaneously entered into the models, only comorbidity (P=0.037) and physical health (P<0.001, Model 2b), or greater disability (P<0.001, Model 2c), or greater WOMAC scores (P=0.001, Model 2d) were associated with false-negative.
  

This study compared the self-reported diagnosed arthritis and musculoskeletal signs and symptoms, suggesting arthritis in a sample of geographically diverse older Australian women. Prevalence estimates based on the two case-definitions of arthritis were not significantly different, but Cohen’s kappa indicated that their agreement was only moderate. While two-fifths (91/223) of the self-reported arthritis cases did not have musculoskeletal signs and symptoms, two-thirds (132/198) of cases identified by signs and symptoms also reported diagnosed arthritis.

The results of study also indicate that self-reported diagnosed arthritis has moderate sensitivity and specificity when using musculoskeletal signs and symptoms as the reference standard. These results are somewhat different from previous studies that included both women and men. One of the USA study involving an older sample from Georgia and using rheumatologists’ summary assessment as the reference standard, found that selfreported arthritis had substantial agreement with the reference standard. This study provides important information about the accuracy of selfreported diagnosed arthritis in a population most affected by arthritis (i.e. older women). Concurrently, the sample represents a strength of this study because: a) survey participants were randomly drawn from an ALSWH cohort which is geographically diverse and b) the response rate in this study was very high. These factors contribute positively to both the external and internal validity of the findings.