Translation and adaptation of the stroke-specific quality of life scale into Swahili

Background Stroke care requires a patient-centred, evidence-based and culturally appropriate approach for better patient clinical outcomes. Quality of life necessitates precise measuring using health-related quality measures that are self-reported and language appropriate. However, most of the self-reported measures were devised in Europe and therefore not considered contextually appropriate in other settings, more so in Africa. Objectives Our study aimed to produce a Swahili version by translating and adapting the stroke-specific quality of life (SSQOL) scale among people with stroke in Kenya. Method We used a questionnaire translation and cross-cultural adaptation. The pre-validation sample of 36 adult participants was drawn from 40 registered people with stroke, from the Stroke Association of Kenya (SAoK). Quantitative data were collected using both English and Swahili versions of the SSQOL scale. The mean, standard deviation (s.d.) and overall scores were calculated and are presented in tables. Results The back translation revealed a few inconsistencies. Minor semantic and equivalence alterations were done in the vision, mood, self-care, upper extremity function and mobility domains by the expert review committee. Respondents indicated that all questions were well-understood and captured. The stroke onset mean age was 53.69 years and the standard deviation was 14.05. Conclusion The translated version of the Swahili SSQOL questionnaire is comprehensible and well-adapted to the Swahili-speaking population. Clinical implication The SSQOL has the potential to be a useful outcome measure for use in Swahili-speaking patients with stroke.


Introduction
The deteriorating health-related quality of life (HRQOL) burden in the stroke population is increasing globally, raising public health concerns (Muralidharan et al. 2019; Shakya et al. 2019). Medical advances in the management of patients with stroke and demographic changes witnessed mainly in populations of lower-middle income countries have led to a decrease in stroke deaths (6.5 million), an increase in stroke survivors (25.7 million) and disability-adjusted life years (DALYs) (113 million) (Donkor 2018).
Stroke has been and remains a major cause of adult activity limitation and participation restriction, which poses a challenge to the HRQOL among people with stroke (Donkor 2018; Gorelick 2019; Odetunde, Akinpelu & Odole 2017). Stroke prevalence is estimated at 0.5% globally with lowermiddle-income countries recording high numbers of stroke deaths and DALYs (4.85m and 91.4m) as compared to high-income countries (1.6m and 21.5m) ( The rising global burden of poor HRQOL among people with stroke has seen the emergence of efforts to integrate precise and valid patient-centred outcome measures to enhance rehabilitation and as such improve stroke outcomes. Perhaps that is the reason why people with stroke seek not only good medical and surgical care but also quality rehabilitation services ( Background: Stroke care requires a patient-centred, evidence-based and culturally appropriate approach for better patient clinical outcomes. Quality of life necessitates precise measuring using health-related quality measures that are self-reported and language appropriate. However, most of the self-reported measures were devised in Europe and therefore not considered contextually appropriate in other settings, more so in Africa. are subjective as they quantify outcomes of an individual's everyday life in their setting to inform rehabilitation plans (Gbiri &  This has been associated with limited knowledge of the importance of PROMs and also the limited existence of contextually appropriate measures. Again, the large number of PROMs that exist has been developed and tested in high-income countries such as the United Kingdom, Australia and America (Hall et al. 2018). Notably, the culture and language in these countries are different from that of Africa in relation to disease expression (Beaton et al. 2002;Guillemin, Bombardier & Beaton 1993). This, therefore, restricts the utilisation of PROMS in multi-ethnic and multicultural contexts (Odetunde et al. 2018), like that of Kenya. Global sub-cultural populations are typically defined in terms of lifestyle and dialect and hence the complexity of direct administration of PROMs developed elsewhere in other contexts (Beaton et al. 2002;Tawa et al. 2021). Either way, modern healthcare practice is appealing for a global response to cultural adaptation of PROMs to enhance the quality of care post-stroke (Hawkins et al. 2020).
Cultural adaptation as a process ensures a clear understanding of self-reported clinical questionnaires by the consumers (Beaton et al. 2002;Hall et al. 2018). The adaptation is not limited to translation to another language but also considers the contextual, semantic, item and conceptual equivalence. The contextual equivalence ensures the words used in the translated version echo the targeted culture, whereas semantics maintains the meaning of words in the translated version as in the original version. Item equivalence ensures translated documents measure the same construct as the source language and conceptual equivalence warrant good content capturing (Hall et al. 2018;Hawkins et al. 2020). The PROM that is the focus of our study is the original English stroke-specific quality of life (SSQOL) scale that was devised over two decades ago by William et al. (1999). The SSQOL scale precisely measures the quality of life of people who have been diagnosed with a stroke and possibly the reason for its comprehensive utilisation as evidenced in its diverse translations into more than 20 languages (Muralidharan et Shakya et al. 2019) have revealed that lower-middle-income countries are the most affected by stroke, thus the need for utilisation of HRQOL measures that are language-appropriate for better outcomes. Therefore, translation and adaptation of the SSQOL scale into Swahili is fundamental in the Kenyan context as well as the entire sub-Saharan Africa to fill the gap. This was in line with scenario 'E' of the cross-cultural adaptation scenarios described by Beaton et al. (2002). Scenario 'E' states that a questionnaire can be adapted to be used in another country and in another language as in the case of our study. Thus, this is the reason for our study that has translated and adapted the SSQOL scale into Swahili.
Our study's specific objectives were:  (Odetunde et al. 2018). The two linguists each generated a written report of the process and two translated documents of Swahili 1 (S-1) and Swahili 2 (S-2).
The two Swahili documents (S-1 and S-2) were then reconciled, as recommended in step two of the guidelines (Beaton et al. 2000(Beaton et al. , 2002Ganvir et al. 2018) by the two linguists and a senior physiotherapist from Kenyatta national teaching, research and referral hospital. The senior physiotherapist was conversant with translation studies, and he, therefore, acted as a scribe to guarantee the semantic and conceptual equivalence of the reconciled document (Odetunde et al. 2020). All items and specific words that were not in line with the original words as used in the scale were acknowledged clearly and a resolution was sought through consensus between the two linguists and the senior physiotherapist (Table 1). This reconciliation step produced one translated Swahili version questionnaire called Swahili 1 and 2 (S-1 and S-2) as recommended by Beaton et al. (2002).
The reconciled S-1 and S-2 document was again back translated to English in the third step to ensure the message in the two versions; that is, the English and the translated Swahili (S-1&2) versions of the SSQOL scale were maintained and corresponding (Odetunde et al. 2018). As in the previous step, the back translation was also accomplished by two different consenting English linguists from a leading publisher (Kenya) although both were naïve to the forward translation process.
The two linguists generated their own written reports of the process and provided one document each, of back translation Swahili one (BTS-1) and back translation Swahili two (BTS-2), respectively, as recommended by Beaton et al. (2002). Further, the two linguists were joined by the first author (acting as a linguist) and together they analysed the inconsistencies in the two documents and by consensus resolved them appropriately producing one version of the back translation of Swahili 1 and 2 (BTS-1 and BTS-2) (Hawkins et al. 2020).

Cultural adaptation process
The three documents, that is, the Swahili version (S-1 and S-2), the back translated version (BTS-1 and BTS-2) and the original English version of the SSQOL questionnaires, were subjected to step IV, for the cultural adaptation process referred to by a committee of experts (

Pre-validation of the pre-final Swahili version
Finally, the PFSV and the original English version of the SSQOL questionnaires were administered to eligible people with stroke for pre-validation purposes (Pedersen et al. 2018). This was the ultimate stage of the translation and adaptation process that focused on guaranteeing the content and face validity of the PFSV questionnaire as recommended by Beaton et al. (2002).
Data were collected from the Stroke Association of Kenya (SAoK) situated within the Nairobi Metropolis. The SAoK is a registered organisation with an aim of rebuilding people with stroke by offering outpatient rehabilitation and wellness services to people with stroke. The association has more than 100 members with stroke, but only about 40 of them are appropriately registered with the association. T-S1: Jagi There was no kiswahili word that could fit the word 'jar' The word 'jar' was translated well to 'jug' T-S2: Birika Therefore, all 40 registered members with stroke formed the study population.

Data collection
Data were collected by the first author and a research assistant using both the SSQOL English (E) and SSQOL Swahili (S) versions of the questionnaires. The Swahili version was assigned odd numbers that is number 1, 3, 5 and so forth while the English version was assigned even numbers that is number 2, 4, 6 in that order, respectively. All the patients at the waiting bay were briefed orally concerning our study and informed that it was voluntary.
Participants were also clearly informed that all the information given would be confidential (Odetunde et al. 2020). Thereafter, the files of those who volunteered to participate were checked to ascertain eligibility. Participants in the acute and severe stages of stroke, and those with a stroke diagnosis of 1 month or less as well as those with difficulties in reading, writing and understanding English or Swahili were not included.
Like in most studies (Odetunde et al. 2018(Odetunde et al. , 2020, eligible participants diagnosed with either ischaemic or haemorrhagic stroke were given a written explanation about our study and signed written consent prior to completing the questionnaire. Consenting adult participants were then recruited systematically after receiving physiotherapy services. Each participant was issued with one questionnaire such that the first participant was issued with questionnaire number 1, which was SSQOL Swahili version and the second participant received questionnaire number 2, which was SSQOL English version. Then the third participant was issued with questionnaire number 3, which was SSQOL Swahili version and so forth, until the last questionnaire was issued as recommended by Odetunde et al. (2020). Participants completed the questionnaires in a private room that was set aside for the exercise. This was achieved by giving participants temporary research numbers for identification purpose.
Further, each respondent was taken through a face-to-face cognitive interview immediately after completing the questionnaire. The purpose of the interview was to assess the relevance, clarity and ambiguity of the items, as well as any important questions that may have been omitted and needed to be included as recommended by Bowden and Fox-Rushby (2003) and Goerman and Caspar (2010). All concerns raised were investigated and captured in the second meeting of experts producing a final Swahili version (FSV) as recommended by Beaton et al. (2002).

Variables
The

Findings of the translation process
The reconciliation of Swahili 1 and 2 documents had a few words that were difficult to translate, and other words had different meanings but were translated as similar. The reconciling team re-investigated those words and phrases and by consensus resolved them as summarised in Table 1.

Expert committee findings
The experts' first committee meeting noted that a good number of the items in the Swahili translation of SSQOL questionnaire were relevant and well-captured as they appeared in the English SSQOL scale. This is also the case in other translations (Odetunde et al. 2018(Odetunde et al. , 2020. On the other hand, back translation indicated some discrepancies of 15 words/phrases/statements to the original English SSQOL scale. The discrepancies ranged from omissions, replacement and contextual inappropriateness to the Kenyan culture. Therefore, they were corrected at the meeting, and a PFSV was produced as summarised in Table 2.

Pre-validation and face-to-face cognitive debriefing findings
The PFSV was tested among the 32 adults with stroke who agreed to be in our study and comprised of 16 males and 16 females who were either good at English or Swahili. The face-to-face cognitive debriefing interview revealed that 46 items out of the total 49 were well-understood in the PFSV and that they were relevant and well-captured. Of the three that were not relevant was item number 1 of the self-care domain, which read 'Did you need help in preparing food?' translated as 'Je ulihitaji usaidizi kuandaa chakula?'. The item was not applicable to two participants that is one male and one female. The reason given by the male participant is that, as a man he never prepared food as it was the role of his wife. The female participant as well, explained that since her childhood, she has never participated in household chores such as cooking.  Also, item number 4 in the mood domain that read, 'I had little confidence in myself' was not clear to most of the participants. Lastly, item number 2 of the vision domain that enquired 'did you have trouble reaching things because of poor eyesight?' was reported to be ambiguous by participants. As a result, the second expert committee meeting was confined to address the concerns. As such, the committee deliberated on the three items, and it was agreed that item number 1 of the self-care domain be maintained as it was to consider the young generation, which is the majority and most urbanised. months or less previously, and 72% had suffered a stroke more than 16 months previously. Respondents working and in business after the stroke were 28% while 72% were not employed or in business. Details are summarised in Table 3.
Respondents' mean scores on the vision domain was high but lower in the energy, mobility, self-care, social roles, upper       (Table 4) Respondents' mean overall total score on the SSQOL(E) was 82, s.d. of 31.04 and SSQOL(S) was 81, s.d. of 20.24 (Table 5).

Discussion
There Like in most studies, the reconciling of the two Swahili documents produced in the forward translation process revealed that some words such as 'mobility', 'upper extremity function' and 'jar' were difficult to translate and therefore had to be borrowed. For instance, mobility was written as 'kusongesha mwili' (moving the body), upper extremity function 'matumizi ya mikono' (hand function) and jar as 'chupa' (bottle). The finding is similar to that of SSQOL Igbo version where extremity was translated as 'ability to use a hand' and jar as 'bottle cover' (Odetunde et al. 2020 Also, in the language domain, item number 2 that read 'Did you have trouble speaking clearly enough to use the telephone?', the words 'clearly enough' were also omitted in both translations and 'telephone' replaced with mobile phone. This was like item number 5 of the mood domain, which read 'I was not interested in food' and back translated as 'I didn't have appetite'. This was an acceptable change as the experiential and cultural equivalent was maintained following the translation guidelines by Guillemin et al. (1993) and Hall et al. (2018) of finding substitute words that better fit the setting.
Besides the omissions, some items lacked consistency in the backward translation. For instance, item number 3 of the mobility domain, whereby the word 'stairs' translated as 'ngazi' was back translated as 'ladder', which was not only inconsistent with the original English but also altered the concept of interest. This was, therefore resolved after a debate by the expert committee members by adding the word house before stairs so as to be consistent and concept appropriate. This finding agreed with a translation study by Wagner et al. (1999) where the word 'Mlima' (hill) was adopted to mean stairs.
Further, some aspects of some phrases also had certain discrepancies and more importantly lacked contextual appropriateness. For instance, item number 2 of the self-care domain that read 'Did you need help eating? For example, cutting food or preparing food?' forward translated as 'Je, ulihitaji usaidizi kula? Kwa mfano, kuchukua chakula kutoka kwa sahani au kuweka chakula kwenye mdomo?' and back translated as 'Did you need help eating? For example, taking food from the plate and putting food in the mouth?' The back translation aspect reading 'taking food from the plate or putting food in the mouth' was not in line with the source language. Similarly, the part reading 'preparing food?' was not in line with eating as preparing food was not a component of eating therefore rendering the phrase irrelevant, and even so, preparing food was already captured in item number 1 of the same domain and thus treated as a repetition. Again, the facet of 'cutting food' was not appropriate to the Kenyan context as it is not an everyday practice in Kenya to cut food with a knife; instead, people use their hands or spoons to take food from the plate. As such, the phrase was revised to 'taking food from the plate or putting food in the mouth'. This was similar to previous translations (Odetunde et al. 2018(Odetunde et al. , 2020 and also in line with Guillemin et al.'s (1993) recommendations of altering phrases to fit the concept and context of application.
In addition, the pre-testing findings revealed that three items were irrelevant, ambiguous and lacked clarity. For instance, item number 1 of the self-care domain that enquired about having trouble preparing food was irrelevant to one male and one female participant. The male participant purported that cooking was the role of his wife while the female participant reported that she had since childhood never participated in performing household chores. The finding partially agreed with that of Odetunde et al. (2018Odetunde et al. ( , 2020 where some men in Hausa and Igbo also argued that food preparation was women's work. However, the committee of experts was reluctant to modify this item because this alteration may restrict its application to the young generation who are the majority and most urbanised. These findings concurred with those of a recent study by Wayessa et al. (2022). Item number 2 of the vision domain was also found to be ambiguous in that most respondents did not clearly understand the meaning. This finding agreed with those by Pedersen et al. (2018) that reported the ambiguity of the same item.

Strengths and limitations
The fact that most of the 42 ethnic communities of Kenya speak Swahili made the translation and adaptation process of SSQOL questionnaire into Swahili easy as there was ready and adequate information on the language that underpinned the achievement of the desired concept. However, there are limited HRQOL tools devised or culturally adapted and valid in Kenya that could be used to correlate with the responses to the Swahili questionnaire. Our study, therefore, had to rely on the original English of the SSQOL scale for comparability, which has not been contextualised in a Kenyan setting. In addition, studies of cross-cultural adaptation are new knowledge in Kenya; therefore, it was challenging to include sector-specific translators who are well-versed in translation.

Recommendations
Further research should be conducted to investigate the reliability and validity of this SSQOL scale in Swahili.

Conclusion
The original English SSQOL scale was translated and adapted to Swahili. The FSV of the SSQOL scale demonstrated the same constructs as the original English SSQOL scale. All items in the 12 domains were understood and captured well with the semantic and experiential equivalence fitting that of the Kenyan culture thus achieving content and face validity of the Swahili version.