Validation of the English Severe Respiratory Insufficiency Questionnaire

RESULTS

A total of 152 patients (75 females) completed the questionnaires (table 1) with 142 and 10 patients receiving NIV and invasive (tracheostomy) ventilation, respectively. Patients with NMD were younger (p<0.0001) and had the highest overnight NIV use (p<0.0001) when compared to COPD, RCWD and OHS patients. There were no differences amongst the other three groups (COPD, RCWD and OHS).

Table 2 shows the distribution of the different conditions leading to chronic respiratory failure. Of the 29 patients with OHS, five patients had lone OHS; seven patients also had COPD; 15 patients had combined obstructive sleep apnoea and OHS; one patient also had scoliosis and one patient had Prader–Willi syndrome. 31 patients were receiving long-term oxygen therapy in addition to HMV; 10 patients each in the COPD, RCWD and OHS groups, and one patient with cystic fibrosis.

Descriptive data of both the SRI and the SF-36 questionnaires are provided in table 3. Five of the eight scales (subscales and summary scale) were normally distributed; in three subscales (SRI-SR, SRI-AX and SRI-SF) normality was formally not achieved. However, the distribution of all SRI scales was rather homogenous as shown by the percentiles given in table 3. Thereby the SRI covered a broad range of the questionnaire's scaling range. The mean score of the SRI-SS (55.9±18.9) was roughly in the middle of the questionnaire’s scaling range, which was also true for most of the subscales (table 3). In contrast, none of the SF-36 scales was normally distributed, and the distribution was severely heterogeneous. This was particularly true for SF-36-PF, SF-36-RP and SF-36-GH, where most patients demonstrated the most severely reduced scores (median ranging between 0 and 30).

DISCUSSION

The SRI is a disease-specific questionnaire with established psychometric properties, which is currently becoming an international standard tool to assess HRQL in patients with severe chronic respiratory failure. The current study establishes that the English SRI, which resulted from professional translation and back-translation of the original German version, has high psychometric properties. Therefore, the English SRI qualifies for being used in clinical studies assessing HRQL in patients with severe chronic respiratory failure. The main findings gained by the present study can be summarised as follows.

The internal consistency as shown by Cronbach's alpha was higher than 0.8 for all subscales except for one (SRI-PF), which, however, was 0.77 and therefore considerably higher than the value of 0.7 considered to be sufficiently reliable for use in group comparisons [23]. Moreover, Cronbach's alpha was 0.93 for the summary scale of the SRI. In the original German version, Cronbach's alpha ranged between 0.73 and 0.79 in three subscales and between 0.80 and 0.89 in four subscales, while it reached 0.89 for the summary scale [11]. This indicates that the reliability of the English SRI is even stronger than the original German version, which, however, is doubtlessly sufficient for use in group comparisons in the research setting [23].

The construct validity of the tool was obtained by factor analytical studies. Explorative factor analysis on the seven subscales using PCA revealed that three subscales loaded on two factors, while one factor could be extracted for the remaining four subscales. Therefore, two subscales could have been theoretically built out of one subscale for the three subscales loading on two factors. In addition, CFA revealed six subscales with two factor loadings; thus 12 subscales could have been theoretically built out of these six scales with the SRI consisting of 13 subscales when adding the remaining subscale with one factor loading according to CFA. As an example SRI-AS, covered both daytime and nocturnal symptoms as reflected by two factors, and this is similar to the recent validation study on the SRI in COPD patients [12]. However, all factors were considerably correlated comparable to the recent validation in COPD patients. Therefore, the original seven-subscale structure was retained, as it was for the recent COPD validation.

Only one factor could be extracted out of the seven subscales for the summary scale of the English SRI, and this is comparable to the two validation studies on the original German SRI [10, 12]. Of note, this factor accounted for 70% of the total variance, but accounted for 59.8% (mixed cohort) and for 58.5% (COPD) of the total variance of the two German validation studies, respectively [10, 12]. For this reason, although high psychometric properties have been established for the original German SRI, the English SRI appears to be even superior. As a consequence, the presence of one summary scale has been established for the English SRI. Since the structure of the SRI contains social, psychological and physical components, this implies that these components cannot be reflected by independent summary scales in patients with severe chronic respiratory failure. This is in contrast to the SF-36, where both physical and mental health factors could be confirmed by factor analysis [4, 21].

The significant concurrent validity between the subscales of the SRI and the SF-36 highlighted the importance of a specific HRQL tool for patients with chronic respiratory failure. The highest correlation was found between the subscales focusing on comparable aspects of HRQL and the least correlation between the subscales which did not. This was comparable to the original validation [25, 26]. SRI-SF strongly correlated with SF-36-VT. Similarly, SRI-WB correlated with SF-36-MH. There was a weak correlation between SRI-AS and SRI-RC with SF-36 scales. This was expected as the SF-36 was not designed to measure sleep disturbances or respiratory complaints, which are reportedly frequent in patients with chronic respiratory failure [25, 26]. SRI-SF showed a moderate-to-strong correlation with several SF-36 scales, indicating that a patient's ability to be involved in social activities is influenced by both physical and mental health factors.

Despite the given validity mentioned above, it has to be pointed out that the translation is formally valid for UK English only. In addition, the SRI is the result of professional forward and backward translation, but a weakness of this study is that this process was not formalised. Nevertheless, British and German researchers serving as authors of this study in addition to two independent native English translators have provided the UK English version of the SRI following a stepwise approach; thus a professional translation has been validated in the current trial. Furthermore, the authors would like to emphasise that for providing further validated translations of other languages, translators need to refer to the original German version and not to the translated English one.

Disease-associated anxiety is an important factor in HRQL in patients on long-term HMV. This is reflected by the scale SRI-AX correlating moderately with SF-36-GH, SF-36-VT, SF-36-MH and SF-36-RE. During the development and validation of psychometric properties of SGRQ and the CRQ anxiety was attributed to psychological disturbances in patients with COPD [6, 27]. Higher degree of anxiety has been shown to be related to higher degree of pulmonary impairment in patients with COPD [28]. Anxieties have been shown to improve after institution of HMV in patients with restrictive lung conditions, COPD and neuromuscular diseases [16, 24]. Duiverman et al. [19] showed that in COPD the SRI total score strongly related to anxiety and depression as assessed by HADS (Hospital Anxiety and Depression Scale).

HRQL in patients with chronic respiratory failure is known to be significantly influenced by the underlying condition [11]. The construct validity is reinforced by the fact that COPD and non-COPD groups could be differentiated by administering SRI. The COPD patients had the worst score for SRI-SS and RCWD patients had the best HRQL, which is in keeping with previous studies with SRI [11, 18].

Previous prospective studies on impact of HMV in chronic hypercapnic COPD have shown conflicting results. Studies by Elliott et al. [29] and Sivasothy et al. [30] did not show significant change in HRQL. However, Meecham Jones et al. [26] and Perrin et al. [31] showed significant HRQL improvement. This difference might be attributed to the fact that different HRQL tools were used in these studies. Neither of these HRQL tools was specifically designed for patients on HMV. Recently, Windisch [16] has shown that specific HRQL as recorded by SRI significantly improved after instituting HMV and this was sustained at 1 yr. It was also shown that overall improvements were similar between COPD, restrictive chest wall diseases and neuromuscular diseases with hypercapnic respiratory failure [16]. Moreover, HRQL changes as assessed by the SRI have even been shown to be dependent on ventilatory strategy in COPD patients, with high-intensity NIV aimed at maximally improving blood gases being superior to the conventional approach of low-intensity NIV, which fails to improve blood gases [13]. This indicates that such a specific tool is better suited to studying this population and it is highly sensitive to change. This validation study paves the way for further studies in this area, especially in the large English-speaking populations using this disease-specific HRQL tool. Given the high psychometric properties of this version there is no reason to believe that this version would not be sensitive to change.

In conclusion, the English SRI is the result of a professional translation and back-translation of the well-validated original German SRI. The present validation study on the English SRI establishes high psychometric properties, which were even higher than in the original German validation study. Other adaptations of the SRI have shown similar high psychometric properties. Thereby, high internal consistency reliability as assessed by Cronbach's alpha coefficient, high construct validity as assessed by both explorative and confirmatory factor analysis, and high concurrent validity as assessed by the correlation analysis between SRI and SF-36 scales have been established. In addition, the English SRI is capable of differentiating between different conditions causing chronic respiratory failure. Other disease-specific questionnaires like CRQ or the SGRQ have not been validated in patients on HMV with varied underlying causes. They do not address all the aspects of daily life that might be affected by HMV and chronic respiratory failure [32]. The English SRI qualifies for being used in researches assessing health-related quality of life in patients with chronic respiratory failure on home ventilation.