Validation of clinical questionnaires is fundamental as such the clinical significance of the tool is accredited.1 Validation of cross-culturally adapted tools have been reported in diverse cultures globally implying the practice is good and evidence-based.2, 3, 4, 5, 6, 7, 8 Even so, research has demonstrated a global increase in utilization of precise and valid patient-centred outcome measures to enhance rehabilitation and improve stroke outcomes.9, 10, 11 Validation of the Swahili version of the SSQOL scale was timely in Kenya as well as the Sub-Saharan Africa and internationally among the Swahili-stroke population due to the fact that Swahili is a national and official language in the aforementioned countries.12

Stroke is the third-leading cause of death and disability (5.7%) of the total DALYS globally, where Low-Middle income countries (LMINs) have recorded an increase rate of 3.7 stroke-related DALYs.13 This is further confirmed by World Health Organization findings that revealed over 16% of people worldwide have a disability whose major cause is stroke. 14 Thus, the more the burden of stroke, the higher the prevalence of disability and the lower the quality of life. 11

Poor health-related quality of life (HRQOL) among people with stroke is of great concern and improving quality of life is thus a key outcome of rehabilitation requiring valid HRQOL measures to quantify outcome.9, 10 The HRQOL measures are patient-reported and so the need to validate them to not only globally but at the local level so as to verify their validity and reliability.1, 15

Validity is about the soundness of a tool to determine its ability to measure what it is envisioned to measure whereas reliability looks at the integrity of a tool in producing similar results if applied in diverse settings.16, 13, 1 Psychometric evaluation is a robust process, in which ensuring adequate sample size is important to correctly power the statistical analysis.17

The original English Stroke-Specific Quality of Life scale (SSQOL) was devised in Europe. 18 The questionnaire has been translated and validated locally and globally to many languages like Marathi, Norwitchian, Arabic, Yoruba, Hausa, Igbo, and Amharic.2, 3, 4, 5, 6, 7, 8 The Swahili version of the stroke-specific quality of life scale is one of the translated and adapted versions in Kenya for the Swahili stroke population however it has not been validated.19 This therefore limits its significance and international utilization, thus the purpose for this study to determine the psychometric properties of the SSQOL Swahili version including- construct, discriminant and known group validity, internal consistency and test re-test reliability.

This study employed a cross-sectional descriptive study utilizing quantitative methods. The Swahili SSQOL is a disease-specific patient-reported outcome measure used to assess the quality of life post stroke. 18, 19 The SSQOL Swahili version has 49 items clustered into 12 domains of upper extremity function, mobility, thinking, language, personality, work, family roles, self-care, social roles, vision, mood, and energy as in the original English version. 19 Each domain has a number of items ranging between 3 and 6 whose responses are in form of a Likert scale. The maximum overall score of the questionnaire is 245 whereas the minimum is 49. A score of 60% of the maximum score denotes a low quality of life. 20

This study aimed to find out the validity and reliability of the final Swahili version of the stroke-specific quality of life (SSQOL) scale among people with stroke in Kenya. The study explored on the construct, discriminant and known-group validity, internal consistency, and test-retest reliability of the final Swahili Version as in other previous studies.2, 3, 4, 6, 7, 8, 5 Mean scores of the SSQOL scale English and Swahili version had no major differences that is 90.17+25.53 and 92.14+28.891 respectively. Same constructs of the SSQOL Swahili version and English version were demonstrated in 10 out of 12 domains. There was a perfect correlation in upper extremity domain (r= 1.00) and the rest of the domains demonstrated a weak correlation (r=0.007-1.00). Item-domain correlations findings showed that items were comparable with their respective domains (r = > 0.7 - > 0.93). Known group validity by gender and age showed no statistically significant differences in all the domain scores of males and females. Whereas, the distribution of domains was the same across age groups except in the language and self-care domains. Internal consistency was good (Chronbach’s α=0.84) and test re-test reliability was excellent (ICC = 0.93).

Our cohort of participants with stroke were similar in the age and type of stroke to other studies conducted in Africa and worldwide.9, 21, 3, 4, 11, 5, 22 However, in age there was a notable number of young adults with stroke ranging between age 18-49 (43%) in our study supporting findings of previous studies that anticipated an increase of stroke in Sub-Saharan Africa among the young adults.23, 24 By implication, the findings support a projection of an increase of 68% of stroke in young generation by 2025.25 Also, in stroke, the findings were different in the body side affected where high income countries and some regional studies revealed the right-side body weakness as the most common as compared to this study that demonstrated left side weakness.3 Also, in this study, female’s participants were more as in other studies conducted in Kenya.21, 26 However, the findings contrast those of similar studies outside Kenya that showed more males to be affected as compared to females.2, 3, 4, 23, 6, 8, 5

Evidence of same construct was demonstrated between the FSV and English version of the SSQOL scale. This is supported by the findings of this study which showed no major differences between the overall mean scores of the English version and that of the Swahili version. These findings agreed with other studies that showed no major differences between overall mean scores of Igbo, Hausa SSQOL and English SSQOL. 8, 18 In addition, the domain scores of 10 out of 12 Swahili SSQOL and English SSQOL demonstrated no statistically significant difference between the scores. Similar findings were also demonstrated in the validation of Hausa and Igbo versions of SSQOL whereby ten out of 12 domains scores were similar. 8, 18 Lack of differences in this study could however be deduced in many ways and imply that the FSV provides a valid evidence of same construct as the English version of SSQOL scale or does not provide valid evidence. The differences could be attributed to the cultural differences that were recognized between the English version and Swahili version. For instance, in the mood domain the English version used the phrase “I was not interested in food” and the Swahili version used “I did not have appetite”.19

The difference in the personality domain may be elucidated to the fact that personality is a concept of human behavior which is diverse across cultures. These findings are similar to a study conducted in Ethiopia that also demonstrated significant differences in the personality domain. 8 Further, in this study, there was a perfect correlation only in the upper extremity function domain and most domains showed no or weak correlation (r = 0.007-1.00). These findings contrast those of similar studies that demonstrated strong correlation for instance, the SSQOL English was correlated with stroke specific Hausa version (r= 0.70-0.92) and SSQOL Igbo version (r=0.58 to 0.87). The different results could be due to the bigger sample size (126) that was used in this study as compared to the Hausa study (86) and Igbo study (50).3, 4 Alternatively, the test may have been underpowered to show significance. The significant correlation of the upper extremity function may be associated to the findings of this study which showed that majority (68%) of the participants had a left sided body weakness and a right hand dominance (92%). This therefore may imply that the participants were able to perform most of the hand function well since the right side of most participants was not affected. The significant correlation may also mean that the translated Swahili version on the upper extremity function domain was valid. The poor correlation in the majority of the domains in this study may be attributed to the fact that the original SSQOL English version that was correlated with the Swahili version is not contextualized into the Kenyan setting therefore it may not have been clearly understood by the participants leading to inconsistency affecting correlation outcome.

This weaker correlation findings perhaps are the reason as to why majority of similar studies opted to correlate the translated versions using measures that have been contextualized into their settings. For instance, the SSQOL Yoruba version was correlated with the Yoruba version of WHOQoL-BRE (r= 0.214-0.360) and SSQOL Amharic was correlated with SF-36 Amharic version (r = 0.21-0.61).3, 4 Also cultural differences that were recognized between the English version and Swahili version in the majority of the domains including- mobility, vision, energy, family roles and language could be the reason as to why there was a weaker correlation.19 The poor correlation may also mean that those domains in the Swahili version are not valid.

The known group validity findings of this study revealed that there were no statistically significant differences between gender in all the domains and overall scores of the FSV. However, the test may be underpowered to detect a difference.16 The findings are consistent with those reported for the Yoruba, Hausa, Igbo, and Amharic version which also reported lack of significant differences across the gender.3, 4, 8, 5 Further, the current study also revealed an association between domain scores of FSV and age except in the language and self-care domains that deferred significantly across the age groups. The findings were similar to the Hausa version that reported statistically significant differences in some domains like language and self-care age groups. 4 The current study findings may be attributed to the fact that the participants aged 70 years and above were the majority affected with stroke and the fact that increase in age leads to a decline in performance could therefore explain why the self-care domain was affected. Further, the findings projected the lowest mean scores on those two domains in people aged above 70 years as compared to other ages further explaining the significant differences demonstrated in the language and self-care domains. Again, the findings of this study demonstrated a significant difference in the older age groups as compared to the younger age groups. The difference may be attributed to the decline of function that is witnessed in old people. The highest mean scores were recorded in younger age groups (except mood domain) and lowest in the oldest age group (except mood, personality and thinking) an indication that psychological functions are less affected across the age groups. The findings a with those of previous studies which also revealed less effects on psychological functions across all age groups affected with stroke.27, 3 These findings therefore imply that age in people with stroke is a strong determinant of health-related quality of life.

Item domain correlations in this study demonstrated a high correlation levels of items with their own domains (r = > 0.7 - > 0.93) than other domains evaluating different concepts. These findings of strong correlations of items with their domains is similar with reports of Yoruba (r = 0.711–0.920), Hausa (r = 0.803–0.979), and Igbo (r=0.58 to 0.87). 3, 4, 5 This further explains the relevance of items with their corresponding domains and discriminant validity of the Swahili SSQOL scale in this study.

The final Swahili version further demonstrated excellent test-retest reliability both in overall scores and domain scores in an interval of 7 days and excellent internal consistency. Findings of this study are consistent with finding from previous studies on different versions of SSQOL scale. For example, the following versions of SSQOL scale reported excellent test re-test; Hausa, Igbo version and Amharic version.4, 8, 5 The internal consistency findings for this study were good and comparable with previous similar studies too.2, 6, 8, 4 The ceiling and floor effects were not significant in this study as in other similar studies. 3, 4, 8, 5

The Swahili version of the SSQOL scale was validated. The FSV of the SSQOL scale demonstrated same constructs as in the SSQOL English version in ten out of the 12 domains. There was a significant correlation in upper extremity domain and excellent test-retest reliability as well as good internal consistency. Further research is needed to assess the validity of the Swahili version of the SSQOL scale using a bigger sample size to give further insights into the FSV psychometric properties.