Article Text
Abstract
Aim: Up to 50% of patients with systemic sclerosis (SSc) have complaints of dyspnoea. We evaluated the independent contributions of dyspnoea to function and health related quality of life (HRQoL) in SSc and also assessed the contributions of pulmonary hypertension, measured by the pulmonary artery systolic pressure (PASP), and interstitial lung disease, measured by the forced vital capacity (FVC), to dyspnoea.
Methods: We assessed dyspnoea, PASP, FVC, function and HRQoL in a cohort of unselected patients with SSc. Multiple linear regression was used to assess the independent contributions of dyspnoea, PASP and FVC to function and HRQoL, after controlling for possible confounding variables.
Results: A total of 194 patients with mean disease duration of 11.6 years were studied. Dyspnoea was a significant independent predictor of function and HRQoL. A model including age, gender, disease duration, disease severity and dyspnoea explained 33.3%, 10.6%, 39.2% and 29.4% of the variance of the Stanford Health Assessment Questionnaire, the Short-Form 36 (SF-36) mental component summary score, the SF-36 physical component summary score and the World Health Organization Disability Assessment Schedule II. PASP and FVC were significant independent predictors of dyspnoea but only 21.9% of the variance in dyspnoea was explained by age, gender, disease duration, FVC and PASP. The FVC was a significant independent predictor of function and HRQoL.
Conclusion: In an unselected population of SSc patients, dyspnoea is a very important contributor to function and HRQoL. Interstitial lung disease, as measured by the FVC, contributes significantly to the sense of dyspnoea, function and HRQoL in SSc. Pulmonary hypertension, assessed echocardiographically by the PASP, predicts the degree of dyspnoea but not function and HRQoL in SSc.
Statistics from Altmetric.com
Up to 50% of patients with systemic sclerosis (SSc) complain of dyspnoea,1 2 a common but non-specific symptom. Potential causes include interstitial lung disease (ILD), airway disease such as bronchiectasis, cryptogenic organising pneumonia, alveolar haemorrhage, gastroeosophageal reflux and recurrent aspiration due to eosophageal dysmotility. In addition, 15–20% of SSc patients may have pulmonary artery hypertension (PAH),2–7 with or without significant interstitial lung disease,5 8 another possible cause of dyspnoea. SSc patients may also experience dyspnoea from other conditions such as arthritis, obesity, anaemia, and/or deconditioning due to inactivity.2
Self-assessment measures of health status and physiologic indices of disease have been shown to correlate with severity of dyspnoea in SSc patients with lung disease.9 However, the independent contributions of dyspnoea in a large population of SSc patients not preselected for severe lung disease has not been carefully studied. The primary aim of the present study was to evaluate the independent contribution of dyspnoea to function and quality of life in SSc patients and secondarily to assess the contribution of pulmonary artery systolic pressure (PASP), measured echocardiographically, and interstitial lung disease (ILD), assessed by the forced vital capacity (FVC), to dyspnoea.
METHODS
The Canadian Scleroderma Research Group (CSRG) collects prospective data on adult patients with SSc recruited throughout Canada. The diagnosis was confirmed by a participating rheumatologist. Disease duration is measured from the onset of the first non-Raynaud disease manifestation. All participating rheumatologists are requested to obtain echocardiograms and standard pulmonary function tests on all patients in their local hospitals. The PASP was calculated from the echocardiogram as 4×(velocity of tricuspid jet).2 The FVC was performed according to ATS standards10 and was expressed as the percentage of predicted.11 The single breath diffusing capacity for carbon monoxide (DLCO) was calculated and was expressed as the percentage of predicted.12 The examining physician assigned a New York Heart Association (NYHA) class to each patient and completed an 11 point numerical rating scale for disease severity. The skin score was calculated according to the modified Rodnan skin score method (mRSS).13
All patients fill out detailed questionnaires that include questions on dyspnoea derived from the Stanford Health Assessment Questionnaire Disability Index (HAQ-DI) modified for use in SSc.14 This question asks “In the past week, how much have your breathing problems interfered with your daily activities?”. This scale has been validated in SSc.14 However, instead of a visual analogue scale, we used an 11 point numerical rating scale (NRS) anchored on the left with “breathing problems do not limit activities” to “very severe limitation” on the right. To confirm the validity of this scale we also used three questions from the modified version of the Pulmonary Functional Status and Dyspnoea Questionnaire (PFSDQ), the PFSDQ-M, which is a questionnaire evaluating functional status and symptoms (dyspnoea and fatigue) in patients with pulmonary disorders.15 16 We used the three general questions on dyspnoea that assess the severity of dyspnoea with the questions “Indicate (a) how you’ve felt on most days during the past month, (b) how you feel today, and (c) how you feel with most day-to-day activities”. The answers were also recorded using an 11 point NRS with anchors at 0 for “no shortness of breath” and at 10 for “very severe shortness of breath”.
Patients completed the HAQ-DI to assess physical functioning. Measures of health related quality of life (HRQoL) include the Medical Outcomes Study Short Form 36 Version 2 (SF-36) and the World Health Organization Disability Assessment Schedule II (WHODAS II). Scores for the SF-36 are summarised into a physical component summary (PCS) score and a mental component summary (MCS) score and range from 0 (worst) to 100 (best). The WHODAS II is a relatively new HRQoL instrument from the WHO. It was developed to operationalise the core dimensions of the International Classification of Functioning, Disability and Health (ICF), the WHO’s revised classification of functioning and disabilities.17 The WHODAS II is different from other health-related quality of life instruments in that it is based on an international classification system, it is applicable across cultures and it treats all disorders at parity when determining the level of functioning. Furthermore, it assesses some functional and health-related issues that are not addressed in the SF-36, including participation in society and sexuality. It consists of 36 Likert-formatted questions. Scores range from 0 (best) to 100 (worst).
Disease severity was assessed with the scale developed by Medsger18 that has been used in other studies.19–21 A severity score of 0 (normal) to 4 (end-stage) was generated for each of the nine systems. To derive a total disease severity score (DSS), the sum of the individual system scores was computed. However, for the purposes of this study, the pulmonary system was omitted from the total DSS as it includes data from the FVC and the PASP. Thus, the range for the DSS was from 0 to 32. Because in our analyses we wanted to control for general disease severity with the effects of pulmonary disease removed from the disease severity index, we felt that this index would be superior to other indices of severity such as the modified Rodnan skin score13 or physician global assessment of disease severity, both of which we also recorded and that we nonetheless used in sensitivity analyses.
Statistical analysis
Correlation coefficients were performed with Kendall tau b. Multiple linear regression used a forced entry method. A p value <0.05 was considered significant. To address multicolinearity, we checked the tolerance values with collinear statistics. All tolerance values were between 0.805 and 0.984, and all bivariate correlations between variables included in the model were ⩽0.234, indicating that multicolinearity was not an issue. The assumptions of homoscedasticity and normality of residuals were checked with residual plots and quantile–quantile plots.
RESULTS
Patients
There were 413 patients in the Registry at the time of this analysis (table 1). Of those, 195 had complete data for FVC and PASP, and this is the subset for which the following analyses were performed. In all, 90% were female, mean (SD) age was 57.7 (12.3) years, mean duration of disease from the onset of the first non-Raynaud disease manifestation was 11.6 (9.3) years and 85.3% fulfilled the ACR criteria for SSc.22 A total of 53.7% of the patients had limited skin involvement, defined as skin involvement distal to the elbows and knees. The SF-36 scores are scaled such that the population norm is 50 and lower results indicate poorer quality of life. It can be seen that the value for our patients for the SF-36 PCS is quite low although not for the SF-36 MCS. The median value for the WHODAS II in a normal population is between 1 and 2 as the scale is skewed strongly toward normal (lower score represents better quality of life). The median for our population represents moderate disability.
The mean PASP was elevated at 38.6 mmHg (15.0 mmHg). The mean FVC was within the normal range but the mean DLCO was low at 69.9% predicted (22.4%). The proportion of patients with FVC<70% was 18.4% and with PASP >50 mm was 10.5%. The total disease severity score was 8.8 (4.0). The mean dyspnoea score was relatively low, representing mild to moderate dyspnoea. The patients exhibited moderate functional and HRQoL limitations with mean HAQ scores of 0.74 (0.66), SF-36 PCS scores of 37.5 (11.3) and WHODAS II scores of 21.6 (16.5).
Some differences between patients with and without complete data for FVC and PASP were found (table 1). Patients included in our study were older than the others. Their mean PASP was lower than the 21 patients who had a PASP assessment but were not included. Their total disease severity score was lower, as was their WHODAS II score and their HAQ.
Relationship between dyspnoea, function and HRQoL
In bivariate analysis, dyspnoea correlated with function (HAQ Kendall tau b = 0.30, p<0.001) and HRQoL (physical component of the SF-36 Kendall tau b = −0.46, p<0.001, mental component of the SF-36 Kendall tau b = −0.17, p = 0.002 and the WHODAS II Kendall tau b = 0.35, p<0.001).
In multivariate analysis, dyspnoea and disease severity were significant independent predictors of function and HRQoL, except for the mental component of the SF-36 for which only dyspnoea was a significant predictor (table 2).
We also performed hierarchical regression analysis where in step 1 demographic data were entered, in step 2 the disease severity score was entered, and in step 3 dyspnoea was added (data not shown). The variables in step 1 account for 1.4%, 0.6%, 6.4% and 4.6% of the variance in the HAQ, SF-36 MCS, SF-36 PCS and the WHODAS II respectively. The disease severity scores incrementally predicted an additional 18.4%, 2.8%, 7.8% and 11.5% of the variance in the HAQ, SF-36 MCS, SF-36 PCS and the WHODAS II respectively and the dyspnoea scores an additional 15.7%, 10.2%, 27.1% and 15.7% of the variances of the respective outcomes measures.
Since the HAQ, PCS, WHODAS, and dyspnoea VAS distributions were significantly skewed, all regression equations were recalculated using square-root transformed variables but the results did not change substantively with one exception. In the prediction of the square root of HAQ, FVC was significant (p = 0.033), whereas it had not been quite significant in the untransformed version (p = 0.051). To facilitate interpretation of parameters, we have presented variable coefficients only from the untransformed equations.
In sensitivity analyses, bivariate correlations with each of the three PFSDQ-M dyspnoea questions were extremely similar to those with the HAQ-DI dyspnoea scale (data not shown). Multivariate analysis was also repeated with the PFSDQ-M question “Indicate how you feel with most day-to-day activities”, and once again the results were almost identical to those in table 2.
Relationship between PASP, FVC and dyspnoea
The PASP and FVC correlated significantly with dyspnoea measured with the HAQ dyspnoea question (Kendall tau b all p<0.001) (table 3) and all three patient global assessments of dyspnoea on the PFSDQ-M (data not shown).
Multivariate analysis demonstrated that the PASP and the FVC were independent predictors of dyspnoea, after controlling for age, gender and disease duration (table 4).
Relationship between PASP, FVC and function and HRQoL
In bivariate analysis, the PASP and the FVC correlated with function, as measured by the HAQ (table 3). Insofar as HRQoL was concerned, they also correlated with the physical component of the SF-36 and the WHODAS II but not with the mental component of the SF-36 (table 3). The negative correlation with the SF 36 PCS results from the fact that lower scores on the SF 36 represent worse HRQoL.
The contributions of PASP and FVC to function and HRQoL were examined with multiple linear regression (table 5).
Disease duration, age, gender, PASP, FVC and the total DSS excluding the lung component, which is assessed as one system using pulmonary function tests and PASP, were entered as independent variables. The FVC and the DSS were significant variables in the models predicting the HAQ, WHODAS II and the SF-36 PCS, although the adjusted R2 values were low (0.167–0.194). The independent contribution of PASP was non-significant. For the SF-36 MCS, there were no significant predictors when the same variables were entered in the model (data not shown).
In sensitivity analyses, all multivariate analyses were repeated using either the total body skin score or the physician global assessment of disease severity in place of the DSS without the lung component. The contribution of FVC to each model was essentially similar whether MRSS, physician global assessment or DSS without pulmonary component was used. In addition, we repeated the multivariate analyses replacing the DSS with all eight individual non-pulmonary organ disease severity scores and there were no meaningful differences in the parameters related to the variables of interest.
DISCUSSION
The relationship between dyspnoea and function/HRQoL
This study explored the relationships between dyspnoea, function and HRQoL in SSc. We have found that dyspnoea itself is a significant independent predictor of function and HRQoL. We demonstrated this in a multivariate model using disease severity, age, gender, disease severity and dyspnoea as independent variables. The models explained a moderated amount, between 29% and 39%, of the variance in function and HRQoL measured by the HAQ, SF-36 PCS and the WHODAS II. In each instance dyspnoea contributed equally or more to these outcomes than did disease severity outside the lungs. This would imply that dyspnoea is a very important contributor to function and quality of life in patients with SSc. In addition, this is the first time that a new WHO HRQoL measure, the WHODAS II, has been used in SSc and we have shown that dyspnoea is also a predictor of worse HRQoL measured with this instrument. By using two separate HRQoL indicators we have strengthened the robustness of this association with dyspnoea.
Contributors to dyspnoea
We found that the PASP and the FVC contribute independently to the degree of dyspnoea in SSc. However, a model including the PASP and FVC as well as disease duration, age and gender accounts for only 15% of the variance in dyspnoea. In other words, a large portion of the variance of the dyspnoea ratings must be due to factors that we are not measuring. Undoubtedly there are factors other than the objective degree of interstitial and vascular pulmonary disease that account for the ratings patients gave their dyspnoea. Dyspnoea, function and HRQoL measures are self-reported by patients and, as such, can be affected by many factors such as depression, distress, perception of the significance of the illness etc. In other studies there is significant disparity between dyspnoea and lung function.23–25 Exercise programs to improve leg23 26 and inspiratory muscle strength27 have been shown to improve dyspnoea, independent of pulmonary function tests. Body mass index,28 anxiety,29 and even depression30 are also independent contributors.
Relationship between FVC, PASP and function/HRQoL
Controlling for disease severity, the FVC is associated with the HAQ, a measure of function, and the WHODAS II and SF-36, measures of HRQoL. These associations were not very strong and this has been reported in other contexts as well.31 32 The PASP was not associated with these outcomes when controlling for disease severity. It is possible that our study is underpowered to detect the contribution of PASP to function and HRQoL. By contrast, it is also possible that the PASP does not relate to function and HRQoL in a linear fashion and that it is only those who have very high pressures, of whom there were only a few in our study, who experience diminished quality of life from their pulmonary hypertension. We have performed other analyses in which we have created a dichotomous variable from the PASP, dividing patients into those with very high pressures (eg, >50 mmHg) and those with lower pressures. Perhaps because of the small numbers in the higher pressure group we still were unable to show an independent relationship with function and quality of life. Also, because some of the elevation of pulmonary artery pressure in SSc may be secondary to interstitial lung disease, it may be more difficult to detect the independent contribution of PASP to function and HRQoL.
The only other study, to our knowledge, that has carefully assessed the contribution of dyspnoea to HRQoL is that of Khanna et al.9 Studying specifically a subset of SSc patients with active alveolitis and disease ⩽7 years duration, the investigators found significant univariate correlations between various measures of dyspnoea and the HAQ and the SF-36. They did not perform any multivariate analysis to control for disease severity but they did look at dyspnoea as a dichotomous variable and noted that function and HRQoL were more impaired in patients with higher levels of dyspnoea. One study using multivariate analysis did find that dyspnoea contributed independently to the HAQ.33 Our study confirms these results, generalises them to a larger population of patients not preselected for pulmonary disease, demonstrates the value of a new HRQoL questionnaire, the WHODAS II, in such analyses in SSc and expands our understanding of the relative contributions of dyspnoea versus other disease manifestations to function and quality of life. In this sample of non-preselected patients, patient-rated dyspnoea was a much more important predictor of function and QoL compared to FVC and PASP, with adjusted R2 values, including covariates, of approximately 30% across measures of function and QoL compared to 15–20% in the equations with FVC and PASP.
There are some limitations to this study. Although we attempted to obtain pulmonary function tests and echocardiograms on all our patients, this turned out to be difficult to achieve. Over half the patients seen by our research group had not had pulmonary function tests and echocardiograms performed, although many of that group did have the former and a few the latter. We are aware of several reasons that inhibited the participating rheumatologists from obtaining tests. In some cases the local ethics committees specifically stated that such examinations could not be performed by the CSRG rheumatologist, but only by the referring physician whose permission was often difficult to obtain. In many other cases there were recent echocardiograms performed by the referring physicians but in substandard echocardiography laboratories. We were unable to use these results and, for provincial billing reasons, unable to order a second echocardiogram. It seems unlikely, however, that patients who were more ill were studied more extensively because the only differences between those who had both tests performed and those who did not were that the group who had the tests were younger, had lower disease severity scores outside the lungs, had better quality of life measured with the WHODAS II and better function assessed with the HAQ. There was also a lower mean PASP in the studied patients, but only 21 of the non-studied patients had an echocardiogram so it is difficult to ascertain the significance of this finding. We feel that it is likely therefore that the reasons for not obtaining echocardiograms and pulmonary function tests in certain patients was more administrative than related to the presence or absence of a suspicion of pulmonary involvement, and therefore our conclusions about the role of dyspnoea as a contributor to function and quality of life are applicable to the entire cohort. The fact that a previous study of patients that were more ill from a pulmonary standpoint demonstrated similar relations between dyspnoea and HRQoL9 suggests that our findings in a population of slightly less ill SSc patients implies that the relationships that we have found are likely to be at least as strong had more of the patients considered more ill been included as well.
As a measure of dyspnoea we used the dyspnoea question from the modified HAQ.14 We also used three separate dyspnoea questions from a thoroughly validated dyspnoea questionnaire15 16 and found almost exactly the same results, so we feel confident that our dyspnoea question was valid.
For disease severity we used the DSS of Medsger et al.18 This is a relatively new measure without extensive validation, although it has been used in other studies.19–21 The concept of a total score has not been extensively validated but we also performed our analyses using the total body skin score and the physician global assessment of disease severity, and found very similar results. In addition we performed the multivariate analyses with all eight individual severity scores from Medgser et al18 and this did not significantly change the results. This would therefore validate the use of the total DSS score as a covariate. We specifically chose the DSS over these other indices of severity because in our multivariate analyses we wished to remove the effects of pulmonary disease from overall severity, and this would not be possible with the other measures.
On balance we feel that the large sample size that we used, despite the loss of over half the patients in our database, supports the robustness of our analyses.
It is important to emphasise that objective measures of lung function and pulmonary artery pressure only explained a relatively small part of the variance of patient dyspnoea. This has been noted by others23–25 and we are actively investigating other potential contributors to dyspnoea because treatment of potentially reversible factors such as distress, depression or poor physical conditioning may improve this sense of shortness of breath and hence quality of life. Our study population was not preselected for the presence of severe pulmonary disease and in fact only 18.4% had an FVC<70% predicted and 10.5% had a PASP >50 mm.
In summary, we have demonstrated that dyspnoea is a significant contributor to the function and quality of life of SSc patients, even in a group of patients without severe pulmonary disease. This is true for an established HRQoL measure, the SF-36, and now for a new measure, the WHODAS II. Pulmonary hypertension and ILD contribute independently to the patient’s sense of shortness of breath. We have also demonstrated that the degree of interstitial lung disease as assessed by the FVC independently predicts important outcomes such as function and HRQoL. Although the level of estimated pulmonary artery systolic pressure does independently predict the degree of dyspnoea, we were unable to show an independent contribution of the PASP to function or HRQoL.
Treatment that improves or delays the decline of lung function should have a significant effect on dyspnoea and quality of life. This has recently been demonstrated in patients treated with cyclophosphamide for interstitial lung disease.34 35 Reduction of PASP should lessen the patient’s sense of shortness of breath. Further studies are required to explain the large portion of dyspnoea that remains unexplained by objective abnormalities in lung function and pulmonary pressure as the discovery of important modifiable variables may lead to interventions that would improve quality of life in patients with SSc.
Acknowledgments
MB is the director of the Canadian Scleroderma Research Group, which receives grant funding from the Canadian Institutes of Health Research, the Cure Scleroderma Foundation, the Scleroderma Society of Canada, the Ontario Arthritis Society, Actelion Pharmaceuticals, and Pfizer Pharmaceuticals.
REFERENCES
Footnotes
Competing interests: None.