This study presents the results of the first cross-cultural translation, adaptation, and evaluation of the psychometric properties of the SSQoL among the Amharic-speaking post-stroke survivors in Ethiopia. Amharic speakers account for 29.3% of the country’s population, with an estimate of 25 million native speakers and over 4 million second-language speakers of this Amharic [48].
The translation, adaptation, and synthesis of SSQoL-Am was easy, acceptable by Amharic speaking stroke survivors, and is ready for use in the local settings. The evaluation of the acceptability of the SSQoL-Am showed that all the domains were below the threshold reported for floor effects (20%). However, the ceiling effect of the vision domain was higher than the threshold reported [26]. Previous studies in the United States of America (63%), Denmark (63.8%), Norway (62.7%), and Germany (51.9%) reported higher ceiling effects exceeding the recommended 20% which is considered poor [10, 14, 18, 30]. The variance and higher ceiling effects of vision domain between the studies could be attributed to the difference in time elapsed between the stroke event and administration of the tool, timing of the questionnaire implementation, types, severity, and duration of stroke among the participants' population. Further, had the Amharic speaking stroke survivors taken the retest questionnaire after a substantial period, they might have achieved adequate recovery by that time.
In the Amharic version of SSQoL, 10 of the 12 domains exceeded Cronbach’s alpha of 0.8, revealing that the items within the domains measured the same concept. This is similar to the findings reported by the original version by Williams et al and other studies (references). The healthy Cronbach’s alpha of SSQoL-Am explains the adequacy of sample size as discussed by the Muus et al work [14]. The SSQoL-Am displayed good test-retest stability as observed by the correlation between 0.83 and 0.92 on SSQoL-Am domains, which was taken one week after the first test. The Danish and the American version of SSQoL reported test-retest correlations of 0.99 and 0.92, the re-administration period was 1 week and 2 hours, respectively [10, 14]. In contrast, the German version [18] had an assessment interval of 1 year and reported test-retest correlations of 0.69, suggesting that the longer test-retest interval affected the correlation and stability of the tool.
The higher test-retest reliability among our respondents might be related to a possible short recall time of one week. The Bland-Altman plot suggests a good agreement between the measures. The Cronbach’s alpha reported in this study is higher than studies in Saudi Arabia, the United States of America, and Mexico. This could be due to the difference in the sample size. For instance, 245 respondents in this study compared to 147, 34, and 31 stroke survivors in Saudi Arabia [11], the United States of America [10], and Mexico [20], respectively. Thus, the observed difference in Cronbach’s alpha between studies is attributed to items studied, sample size, and heterogeneity of respondents. Further, the present study included stroke survivors from multicenter, with broader chronicity and severity. In contrast, the study in the USA recruited one-month post-stroke subjects. Our results are consistent with the internal consistency of SSQoL reported by the Danish, Brazil, Turkish, and Yoruba version [12, 14, 18, 22].
The SEM and MDC provides an indication of the absolute reliability of a tool and genuine agreement of repeated measurements [49]. The narrow MDC value and small SEM in the present study suggest that SSQoL-Am is a stable measure across repeated administration. Further, the strong reliability, low SEM (< 0.4) scores in most of the domains in the present study indicate that SSQoL-Am is a very reliable tool for clinical and researching HRQOL among stroke survivors. A minor degree of the change in the domain score reflects a true change in the construct, and a change score of 1 SEM might indicate a minimal clinical difference [50]. The domain-wise SEM observed in this study is consistent with the Norwegian version of SSQoL [29].
In this study, the content validity of SSQoL-Am was conducted by a qualitative cognitive debriefing. The construct convergent validity was established by conducting a correlation test of the tool with the sub scales SF-36 tool [51]. The findings of cognitive debriefing suggest that SSQoL-Am content of interest (Health-related QOL) is captured, participants understood the content, and the concept being measured is easy to understand by Amharic speaking Stroke survivors. As hypothesized, the construct convergent validity of SSQoL-Am is reflected in its significant strong to moderate correlation of mobility and self-care domains with the physical functioning sub-scale of SF-36. A similar strong correlation between these domains of SSQL and SF-36 findings have been reported in the original version and other previous studies [10, 11, 29]. This positive correlation may be explained by the relationship between self-care needs and physical functioning abilities in conditions like stroke.
The construct validity scores (r2) on the SSQoL-Am ranged between 0.21 and 0.61 (Table 4), which suggests an adequate linear relationship between most of the SSQoL-Am domains and the compared SF-36 subscales. This finding is similar to the linear relationship reported by Muss et al and Williams et al [10, 14]. But, the social and personality domain of the SSQoL-Am tool had a low linear relationship with the social function and mental health subscales of SF-36, than reported by Williams et al, Odetunde et al, and Muus et al [10, 14, 17]. This difference can be explained by the variance in elapsed time between the stroke event and administration of the questionnaire, longer the recovery time better the social adaptation. Further, the personality domain attempts to predict human behavior, which varies widely across cultures. As noted by Muus et al, the item ‘I had sex less often than I would like’ (Social role domain) had been reported to have the largest missing response. The sensitivity or private nature of items like this could have caused disagreements. The mobility and self-care domains of SSQoL-Am had good agreement with SF-36 (r2 0.57, 0.61 respectively) in this study. These findings are similar to several versions of SSQoL questionnaires with r2 ranging between 0.41 and 0.62, suggesting that SSQoL-Am can measure patient mobility and self-care, similar to the SF-36 tool.
Ethiopians are socially more interrelated and social activities are one of priority of their life (reference). The energy domain had a weak but positive correlation with the SF-36 vitality subscale was 0.35, which is higher than what was reported in Mexico (0.08), and Saudi Arabia (0.25); however, the findings are lower than that reported by the United States of America (0.51) and Denmark (0.52) [10, 11, 14, 20]. This difference probably indicates that the established scale is more specifically attributed to translation in questionnaires, cultural differences, and duration of the condition. The SSQOL-Am total score demonstrated a significantly strong correlation with the total score of SF-36 (rho = 0.74, p < 0.001) which was higher than reported in the original tool developer (rho = 0.65, p < 0.001) (14). The overall higher agreement could be attributed to variations in demographics, absence of homeless, people living in organizations in this study, and test-retest duration. Nevertheless, the items of SSQoL-Am domains individually demonstrated a good level of convergent validity (> 0.50) except for the social domain, which had a relatively weak correlation between items and other domains. The weak correlation could be partly explained by the discriminant validity of the tool.
The CFA of SSQoL-Am did not conform to the proposed theory of two-factor structure. The EFA indicated that SSQoL-Am is a predominately a one-factor and secondary nine-factor structure. The factor loads of items located in each factor were found to be moderate and the variability in the number factors explained in this study and the results from other settings [18, 21] could be attributed to the perceptual differences, and subjective nature of the items in SSQoL. This finding suggests the need for further construct evaluations of the SSQoL-Am in Ethiopia. Nevertheless, more importantly all the items of upper extremity functions and 4 out of 5 items of self-care fell into one subdimension which almost accords conceptually with the subdimension intended by the original version. The sigficance of our factor analysis is that SSQoL-Am should be considered a tool that may assess stroke related QoL with potentially other underlying constructs when used in research or clinical settings in the Ethiopian stroke survivors.
Our study has few limitations. There are higher proportion of ischemic stroke survivors in this study and the different outcomes influenced by the severity of the disease may limit the generalizability of the findings. The chronicity of stroke among participants was significantly different at the baseline. It is understood that the stroke patients at different phases of the post-stroke recovery may report and perceive different HRQoL concerns. It is not clear if the SSQoL-Am demonstrated that validity. Finally, pilot testing and cognitive debriefing were conducted with a small sample. Hence, future studies should address these gaps and the findings of this study should be interpreted with caution based on its limitations.
Conclusion
Impaired quality of life is often the negative consequence of stroke. The SSQoL is a reliable, valid, and very useful tool that can help evaluate patients QOL. The English version of SSQoL was successfully translated and adapted to the Amharic language. The Amharic version of the SSQoL-Am scale is a tool with good psychometric properties and adequately supported construct validity. It appears to be a suitable tool for use in future epidemiological studies and clinical studies among Amharic speaking Stroke survivors.