Patients and methods

Patients

Canadian and Australian patients who fulfilled modified New York criteria for AS, according to their treating rheumatologist and were older than 18 years received a postal questionnaire in 2005 in the context of a research project to assess the burden of AS. The Canadian datasets were acquired through participation of academic and community-based centres in northern Alberta, the Ontario Spondylitis Association and the Arthritis Society, British Columbia division. The Australian datasets were acquired through collaboration of the Austin Spondylitis Clinic in Melbourne, Australia, which receives referrals from both general practitioners (60%) and community rheumatologists (40%). The study protocol was approved by the ethics committees of the respective institutions and all patients provided informed consent for use of the data for scientific purposes.

Assessments

The first part of the questionnaire examined sociodemographic and disease characteristics, including age, sex, ethnicity, marital status, education and employment. For further analyses, the five ethnic groups were combined into Caucasian versus non-Caucasian (Oriental, Asian, Hispanic, native Australian, other) backgrounds. Education level was dichotomised into lower (primary, incomplete or complete high school) and middle (college and technical school) or higher (university) educational levels. Five categories of marital status were grouped into “living with a partner” (married and common law) and “not living with a partner” (single, divorced, widowed). The four employment states were dichotomised into employed (employed/self-employed) and not employed (retired, houseworkers and unemployed).

Patients filled in the date of diagnosis of the disease and completed questionnaires to assess AS-related disease activity (Bath Ankylosing Spondylitis Disease Activity Index (BASDAI); range 0–10; higher values indicating greater disease activity)13 and physical function (Bath Ankylosing Spondylitis Functional Index (BASFI); range 0–10; higher values indicating worse functional status).14 HR-QoL was assessed by a disease-specific (ASQoL)15 and a generic (EQ-5D) instrument.16 The ASQoL comprises 18 questions that cover relevant aspects of QoL, including pain and stiffness, fatigue, mood, physical function and daily activities. Each question can be answered by “yes” or “no” and the total score ranges from 0 to 1 (higher values indicating worse QoL). The EQ-5D includes five dimensions of health: mobility, self-care, usual activities, pain/discomfort and anxiety/depression and can be rated by three degrees of severity scale: “no problem”, “some problems” and “severe problems”. The overall score is calculated by the Dolan transformation function and represents a population preference (utility) for that specific health state. The score ranges from −0.59 (worse than death) to 1.0 (perfect health). Finally, patients completed the Rheumatoid Attitudes Index (RAI) Helplessness Subscale (RAI/Helplessness) that comprises five items, each rated on a five-point Likert scale and evaluates perceived control over the disease (helplessness). For illustration, example items are: “My condition is controlling my life” and “No matter what I do, or how hard I try, I just can’t seem to get relief from my pain”. The score ranges from 0 to 25, higher scores indicating more helplessness.17

Statistical analysis

Descriptive statistics were used to explore differences in demographic and disease characteristics of the samples from the different countries. An independent Student t test, one-way analysis of variance, χ² or Kruskal–Wallis test were used, as appropriate, to examine differences between groups.

To explore the relation between contextual factors (either environmental or personal) in combination with disease-related variables and HR-QoL (ASQoL and EQ-5D), multiple stepwise linear regression analyses were performed. Independent variables were included progressively in three blocks. In the first block, age and gender were included as generic personal factors influencing HR-QoL. In the second block, nationality, marital status and employment status (predominantly environmental factors) and ethnicity, education and helplessness (predominantly personal factors) were added. Interactions between different contextual factors introduced in this block were explored in additional regressions. The interdependence of contextual factors was also explored by adding variables one by one in varying order. To assess the further contribution of variables related to functioning and disability, the model was completed by a third block of variables that included BASFI and BASDAI (well-known determinants of health status), while controlling simultaneously for disease duration. To avoid multicolinearity between the independent variables, a tolerance threshold of <0.2 (indicating >80% variance between some of the variables) was considered unacceptable. The total percentage (R²) and change in the percentage (incremental R²) of explained variance and the partial correlation coefficients after each step are presented. Analyses were performed using SPSS 14.0.

Results

Patient demographics and disease characteristics

Of 600 and 164 questionnaires that were sent to Canadian and Australian patients with AS, 417 (69%) and 105 (64%) questionnaires were completed and returned. Table 1 presents the sociodemographic and disease characteristics of the study groups. For 111 of the 183 Canadian non-responders there was information on some demographic and clinical characteristics over the same period owing to the availability from data from a clinical database to which patients had consented: 68.4% were male, mean (SD) age 51.7 (14.9) years, disease duration 18.3 (13.5) years, BASDAI 4.0 (2.4), BASFI 3.8 (3.5) and ASQoL was 7.0 (6.7), which is fully comparable to the data of the Canadian participants.

Between countries, several differences among demographic and contextual factors were seen. Specifically, Australian patients were younger and more commonly non-Caucasian, less often living with a partner and more often employed. When comparing functioning and disability, Australian patients had shorter disease duration, lower disease activity and better physical functioning. As a result, overall HR-QoL tended to be better in the Australian group, but was not significantly different from that of the Canadian patients.

Relationship contextual factors and HR-QoL

Table 2 shows the results of the regression analyses for ASQoL and EQ-5D, respectively. No multicollinearity between the independent variables was shown as indicated by the tolerance, which was above 0.4 for all variables. The total explained variance of the final model was high, 71% for ASQoL and 55% for EQ-5D. Age and sex accounted for only a small part of the variance in HR-QoL, in which men expressed a somewhat better EQ-5D-assessed QoL. When adding other contextual factors, the explained variance increased to 47% and 37% for the ASQoL and EQ-5D, respectively, and could be mainly explained by the contribution of employment and helplessness and to a lesser extent by ethnic background and education (EQ-5D). An interaction was seen between education and helplessness (p = 0.03), when EQ-5D was the independent variable, and between education and employment when ASQoL was the outcome (p = 0.001). When exploring the relation between helplessness, education and QoL, it was seen that helplessness mediates the association between education and QoL. When adding BASDAI and BASFI in the third block, explained variance increased by 24% and 18% for ASQoL and EQ-5D, respectively. The association between employment and helplessness with HRQoL decreased, especially for helplessness, but remained significant. The low association with education and ethnicity disappeared. Disease duration, as a possible confounder of functioning, did not have a significant effect.

Discussion

This study highlights the role of contextual factors, more specifically of helplessness (a personal factor) and employment (an environmental factor), in self-reported HR-QoL among patients with AS. Patients who are not employed and who perceive more helplessness report worse QoL. The finding is consistent for both generic and disease-specific QoL. No independent effect was seen for nationality or ethnicity, but there was a small additional effect of gender—women had worse QoL (EQ-5D). The important effect of helplessness and employment remained after correcting for variables representing functioning and disability, as measured by BASDAI and BASFI.

The importance of BASDAI and BASFI is not surprising since these instruments share components in their construct that are part of the HR-QoL instruments, such as pain or limitation of physical function (fig 1). The association in cross-sectional studies between employment and HR-QoL has been shown previously.6 However, the role of helplessness, a personal factor, has not been shown previously in AS. Helplessness refers to a psychological state in which patients believe that their efforts to control the disease are ineffective. The concept is based on the Learned Helplessness Theory, which suggests that in the face of unpredictable and uncontrollable events, people can learn to become helpless. Helplessness is closely linked to self-efficacy and both are the counterparts of the spectrum of “controllability”, an important concept in illness representations. To assess helplessness in patient with a rheumatological condition, the 15-item RAI was developed.18 The five-item Helplessness subscale is considered to be conceptually cleaner because the items consist of a single factor and the questionnaire is easier to complete.17 In rheumatoid arthritis, good correlations of the RAI/Helplessness subscale were found with self-perceived pain, anxiety, depression, general rating of limitations, function (Health Assessment Questionnaire), but also with objective assessed disease activity, compliance, self-esteem, information seeking and mortality.17 19 20 21 22 In a 2-year follow-up study, helplessness at baseline was still associated strongly with physical and emotional outcomes.23 One study showed that helplessness predicts the role of disease-modifying antirheumatic drugs.24 As the Learned Helplessness Theory suggests that people become more helpless when confronted with situations that are uncontrollable, it is not surprising that the strength of the association between helplessness and HR-QoL decreased somewhat when adding BASDAI and BASFI to the model. It should be noted that this cross-sectional study cannot help to explore the directionality of the influence between severity of the disease and helplessness.

In our model, an interaction was seen between education and helplessness with generic HR-QoL (EQ-5D) and between education and employment with disease-specific HR-QoL (ASQoL), indicating that helplessness and employment have more impact in people with lower educational level. When exploring the relation between education, helplessness and QoL further it became clear that helplessness partly mediates the association between educational level and QoL. This was previously confirmed in rheumatoid arthritis,19 and is relevant since it could serve as a starting point when trying to reduce the negative effect of poor socioeconomic background on HR-QoL. The literature on inflammatory rheumatological diseases, however, does not provide examples of studies aimed to influence helplessness. Notwithstanding, educational programmes or cognitive education can improve self-efficacy and patient outcomes.25 This suggests that it might be worth to direct such programmes more specifically towards people with more perceived helplessness.

This study specifically aimed to explore the influence of nationality and ethnicity on QoL since the cross-cultural comparability of HR-QoL is of increasing importance.10 26 Both, nationality and ethnicity represent the cultural background of subjects that is determined by (cultural) perceptions of illness, the meaning of HR-QoL, expectations of (health) care, and are important when interpreting perceived HR-QoL. Nationality differs from ethnicity as it also captures the influence of the social security system such as access and quality of healthcare.11 12 Although there were huge differences in sociodemographic and disease characteristics between both countries, neither nationality nor ethnicity had an independent effect on HR-QoL in the final model. This means that in international comparisons, despite similar sampling of patients, case mix should not be neglected.26 Notwithstanding, our findings conflict with the literature, which suggests large differences in HR-QoL across countries.11 Lack of influence of ethnicity and nationality should be interpreted with caution; the subsample of patients from non-Caucasian origin was small and only two nationalities were included which have similar ethnic origins, cultures and organisational structures.

In this study, a selection of all possible explanatory factors of HR-QoL was included. It was aimed to represent all components of the ICF, including the contextual factors. Recently, the ICF Core Set for AS was defined, which confirms the role of all components in functioning for patients with AS (see article on page 102).27 The choice within the components was based on knowledge derived from the literature and on recent HR-QoL research. Although the ICF proved to be useful for this purpose, one of the difficulties was that variables often represent several component. An example is nationality, which represents personal (cultural) as well as environmental (organisational) factors. As such, it was an interesting exercise to reflect on the exact meaning of the variables chosen.

An underestimated consequence of our findings is that HR-QoL is used as a clinical outcome and also used to calculate QALYs, which are important in economic evaluations. Our data suggest that transferability of QALYs across populations might depend on demographic and disease characteristics and also on personal factors (helplessness and employment) and these variables are important when interpreting data on HR-QoL between patient groups.

In summary, this study demonstrates the importance of helplessness and employment status on self-reported health states in patients with AS. It is suggested that helplessness partly mediates the effect of education on QoL. Considering that HR-QoL is also used to calculate QALYs and plays a part in decision-making, the role of personal factors and employment status in HR-QoL should be taken into account.