Interstitial lung diseases (ILDs) are a heterogeneous group of pathologies with similar clinical, radiological, and functional characteristics. The occurrence of ILD varies with prevalence rates ranging from 17/105 to 80/105 and annual incidence rates ranging from 4.1/105 to 30/105.1–3 In Brazil, the incidence of interstitial pulmonary fibrosis is 0.4/10 per year, while the prevalence is 14/10.4 In addition to respiratory problems, patients with ILD show systemic manifestations that directly impact their quality of life.5

Although there is a vast number of generic instruments to assess health-related quality of life, better results are obtained when valid and specific instruments are used for targeted populations.6,7 The St George's Respiratory Questionnaire (SGRQ) was initially developed and validated to assess health-related quality of life in patients with chronic obstructive pulmonary disease (COPD).8,9 The SGRQ has acceptable validity and reliability to use in individuals with idiopathic pulmonary fibrosis.7 However, due to large clinical differences between patients with COPD and those with idiopathic pulmonary fibrosis, a specific English version of this questionnaire for this population, named the SGRQ-I, was developed.6 Although the SGRQ-I was specifically developed for patients with idiopathic pulmonary fibrosis, the clinical and functional similarities among the various subtypes of ILDs allows the use of the SGRQ-I for the broad group of ILDs. In fact, the SGRQ-I has recently been used as a clinical outcome in a randomized controlled trial investigating the effect of exercise training for patients with different ILDs.10

The SGRQ-I has been cross-culturally adapted into Spanish with adequate internal consistency, reliability, and validity.11 Likewise, the SGRQ-I may be valid for Brazilian patients with ILD; however, this has not been investigated. The aim of the present study was to translate and cross-culturally adapt the SGRQ-I to Brazilian-Portuguese and to test its measurement properties (i.e. reliability, validity, minimal detectable change (MDC), and ceiling/floor effects) in patients with ILDs.

Participants' characteristics are described in Table 1. Thirty patients diagnosed with ILD were initially assessed and considered eligible for inclusion in the study (15 men; mean age of 59 ± 10 years; median FVC = 73% [61–80] of the predicted value). The most prevalent diagnosis was idiopathic pulmonary fibrosis (n = 18), followed by pulmonary fibrosis related to diseases of the connective tissue (n = 5), diseases related to inhalation of particles (n = 4), and non-specific interstitial pneumonia (n = 3). Most of the participants were literate (93%) but had a low education level (i.e. 50% of them had only elementary education completed). The cross-cultural adaptation and translation of the questionnaire did not undergo significant changes except for item 4 of Section 5. The original question asks, "My breathing makes it difficult to do things such… play bowls or play golf". These two activities (bowls and golf) were excluded from the item as they were not applicable to Brazilian patients (Supplemental online material).

The median length of time for completing the SGRQ-I during the initial, second, and third visits was 461 s [413–536], 476 s [430–552], and 422 s [368–470], respectively (p = 0.052). The median SGRQ-I total score for the 3 visits was 50 [36–63] points, 48 [38 −70] points, and 56 [40–68] points, respectively (p = 0.94). SGRQ-I domains and total scores for the initial visit are reported in Table 1. The total score of SGRQ-I had excellent intra-assessor (ICC= 0.93; 95%CI: 0.85, 0.97]) and inter-assessor (ICC = 0.88; 95%CI: 0.77, 0.94) reliability. Internal consistency analyses were deemed adequate for total score (Cronbach's α = 0.88), activity (Cronbach's α = 0.83), and impact (Cronbach's α = 0.79) domains but inadequate for symptoms domain (Cronbach's α = 0.48). The SEM for the total score was 6.5 points (or 13%) and the MDC was 12.8 points. The percentage of patients with maximum and minimum score were respectively 3% and 0%, indicating no ceiling and floor effects. Reliability date are provided in Table 2.

For the validity analysis, the total SGRQ-I score showed negative moderate correlation with the physical health component of the SF-36 (r = −0.53; p < 0.05). The full set of correlations between SGRQ-I and the domains of the SF-36 is provided in Table 3. There was no significant correlation between any domain of the SF-36 and SGRQ-I symptoms domains (−0.35≤r ≤ 0.20; p > 0.05 for all). The impact domain of the SGRQ-I correlated with the functional capacity, general health status, and vitality domains of the SF-36 (−0.71 ≤ r ≤ −0.39; p < 0.05 for all).

The SGRQ-I was not able to discriminate health-related quality of life according to the disease severity, age, sex, educational level, and cognitive status (discriminative validity) (Table 4). Finally, the power analysis of the correlation between SGRQ-I total score and physical health component of the SF-36 was 0.83.

SGRQ-I scores significantly correlated with the MRC (symptoms r = 0.42, p = 0.02; activity r = 0.71, p < 0.0001; impact r = 0.62, p = 0.0004; total score r = 0.66, p = 0.0001). There was no significant correlation between SGRQ-I scores and exercise capacity [−0.36 ≤ r ≤ −0.25; p > 0.05 for all], pulmonary function outcomes [−0.03 ≤r ≤ 0.20; p > 0.05 for all], and daily activities of light or moderate intensities (0.16 ≤r ≤ 0.30; p > 0.05 for all). Daily steps correlated significantly with symptoms of the SGRQ-I (r = 0.37, p = 0.04).

In the present study, the translated version of the SGRQ-I demonstrated good to excellent intra- and inter-rater reliability and good to excellent internal consistency. Further, the Brazilian-Portuguese version of the SGRQ-I is valid to be used in patients with ILDs. Consistent with previous work, the reliability analysis in this study demonstrated excellent values of ICC in most domains and total score.6,43 Yet, scores were somewhat higher in the intra-assessor analysis. The difference in total scores between assessors, however, was negligible with a clear overlap of the CIs. Also, in the intra-assessor analysis the scores of the second application were slightly higher, which suggests a learning effect. This, however, is not supported by the statistical analysis which showed adequate ICC values and no statistical difference between the assessments. The values of Cronbach's α confirmed the adequate internal consistency of the SGRQ-I except for symptoms (i.e. α < 0.7). In a systematic review of the psychometric properties of the SGRQ in patients with idiopathic pulmonary fibrosis, Swigris et al.33 also reported low values of internal consistency in the symptoms domain. A possible explanation for this is that questions related to symptom assess respiratory complaints that are not common in ILDs. For instance, items 4 and 5 of the first part of the questionnaire refers to "wheezing" and "respiratory crisis," symptoms more commonly reported by patients with obstructive lung diseases.

To the best of the authors' knowledge, this is the first study investigating the MDC of the SGRQ-I. This statistical approach helps clinicians to identify changes that are not merely a normal variation of the tool but instead changes in the clinical condition of the patient.44 It, however, does not mean that changes on MDC superior to 12.8 points are considered clinically relevant for patients. Future studies are needed to determine whether this cut-off resembles clinically relevant changes in patient's quality of life.

Recently, Prior and colleagues45 demonstrated that less than 15% of patients with ILDs reached highest (celling effect) or lowest (floor effect) deciles of possible scores on the SGRQ-I. Consistent with these results, our findings also did not identify a significant proportion of patients with maximal/minimum scores in the questionnaire (i.e. less than 5% of patients). Albeit these results should be confirmed in larger sample size, this is a good indication that the tool is not limited by construct constraints.

The analysis of the construct validity of the SGRQ-I or Brazilian-Portuguese language presented similar values to those observed in the validation of the English version.6 These results confirm that the translation and the cross-cultural adaptation in the present study did not impact on the essence of the questionnaire. Indeed, there was no need to adapt the text during the translation process significantly. Only item 4 of the fifth section was changed due to environmental and cultural differences, as reported in the results section. Importantly, the questionnaire was designed to be applied in idiopathic pulmonary fibrosis. The present study expands the results of the reliability and validity of the SGRQ-I questionnaire for utilisation in patients with ILD.

The analysis of discriminative validity showed that it was not possible to anticipate a better or worse quality of life based on age, sex, educational level, cognitive status, and the severity of the disease. This finding, however, needs to be interpreted with caution. The sample size of the present investigation was not powered for this analysis, and these results likely change if using a larger sample size. Additional validation analyses were recently published using the English version of the SGRQ-I and confirmed that it was good at discriminating patients with different stages of the disease.45

Similar to Capparelli et al.,11 we also found a significant correlation between SGRQ-I and MRC scores. Conversely, none of the other investigated outcomes presented significant correlations. This is not particularly surprising as the SGRQ-I also failed to show strong correlations of the same outcomes in the validation study for the English language.6 Lubin et al.46 demonstrated that dyspnea was one of the strongest determinants of health-related quality of life in patients with idiopathic pulmonary fibrosis, being superior to disease severity. In other respiratory diseases, a reduced exercise capacity present meaningful relationship with health-related quality of life, especially in more severely affected patients.47 The sample in the present study did not include many patients with marked reduction of exercise capacity or muscle strength. It is possible to hypothesise that the lack of correlation between SGRQ-I and 6MWT is a consequence of the preserved muscle function in the assessed patients.

The results of the present study need to be interpreted in light of some potential limitations. First, the sample size is somewhat small. The low prevalence of patients with ILD and the monocentric design of the present study have limited participant recruitment. However, other authors found acceptable results with a similar sample size.11 But, the post-hoc power analysis of the SGRQ-I validation (power=0.83) confirmed the sample size of the present study did not hamper the interpretation of our results. Second, the questionnaire was applied in patients with different types of ILD, despite the original version being specifically developed for patients with idiopathic pulmonary fibrosis. Although 60% of the study sample was comprised of people with idiopathic pulmonary fibrosis, our results were similar to the previous study on the validity of SGRQ-I.6 Also, investigating the validity of SGRQ-I in specific diseases was beyond the scope of the present study. Third, in the process of translation and cross-cultural adaptation of questionnaires, the backward translation should ideally be performed by two independent researchers. Finally, the SGRQ-I was initially developed to be self-administered. In the current study, questionnaires were researcher-administered. This was done as we anticipated patients with low literacy levels to be included in the present study. Given there is considerable variation of scores when questionnaires are self- versus researcher-administered, we attempted to reduce bias by standardising the application method.31

The Brazilian-Portuguese version of SGRQ-I, albeit conducted in a relatively small sample, seems to have adequate measurement properties justifying its use in the Brazilian population.

None of the authors have conflicts of interest to declare.