Gestational diabetes mellitus (GDM), defined as ‘any degree of glucose intolerance identified at onset or first recognised during pregnancy’, is a major global public health concern.¹ Although the true prevalence of GDM remains unknown because of inconsistent screening methods and diagnostic criteria,² the International Diabetes Federation estimated that the global prevalence of hyperglycaemia in pregnancy is approximately 21.3 million (16.2%).³ Of these, the majority of cases (86.4%) were attributed to GDM.^3,4,5 In low- and middle-income countries where there are limited resources and other socioeconomical and environmental factors that negatively affect health and access to healthcare, the projected burden of GDM is believed to be increased⁶ and under-managed. In South Africa, for instance, a lower-resource and diverse country comprising of a large proportion of populations living in rural and/or remote areas,⁷ the prevalence of GDM was estimated to range between 9.1% and 25.8%.^8,9

If not managed properly, however, any degree of hyperglycaemia first detected in pregnancy, including GDM, is associated with a myriad of complications during and after pregnancy.^10,11 Epidemiological surveillance indicates that GDM mothers and their offspring develop chronic non-communicable diseases (NCDs) easily.^10,11,12 Gestational diabetes mellitus is specifically associated with increased risks of caesarian section, pre-eclampsia, gestational hypertension and type 2 diabetes post-partum for the mother.^10,11,12 For the child, there are risks associated with neonatal hypoglycaemia, macrosomia and developing type 2 diabetes later in life.^10,11,12

Many serious adverse events associated with GDM can however be mitigated through educating individuals about self-managing their condition and modifying their lifestyle choices.^13,14,15 A good comprehension of GDM, adequate lifestyle modification and accurate self-management of their condition have been shown to translate into better glycaemic control, which may lead to reduced peri-natal, post-natal and long-term health complications for both mother and child.^13,14,15 In primary health care settings, GDM patients are typically screened, monitored and managed by primary healthcare nurses and other health professionals. Part of the management of patients with GDM involves the implementation of health education strategies around self-monitoring and self-management of the condition.^13,14,15 In areas where regular access to healthcare is, however, limited, for example, for patients living in remote or rural areas,⁷ these health educational strategies around self-management of the condition between their healthcare visits become even more vital. Educating patients about self-management is, however, applicable not only for patients in South Africa who live in rural or remote areas but also for patients in other countries where access to healthcare is limited because of similar factors.

To apply the most appropriate patient health education strategies and prescribe the most optimal self-management programme, the health professional needs to assess the patient’s baseline understanding of the condition and health literacy.¹⁶ At the time of the study, the authors were aware of only one questionnaire that assessed the knowledge base of patients around GDM, known as the Gestational Diabetes Mellitus Knowledge Questionnaire (GDMKQ). The original GDMKQ was developed in Malaysia by Hussain et al.¹⁷ and combined a number of existing questionnaires to produce a 15-item questionnaire. To our knowledge, however, no similar validated questionnaire fit for purpose in South Africa exists. This said, directly applying an already existing outcome measure that was developed for a different population and culture is not recommended.¹⁸ The use of cultural- and linguistic-inappropriate questionnaires may lead to misinterpretations of questions, inadequate application of the scoring system and anchores and/or references that are culturally not appropriate, which in turn may result in inaccurate measurement.¹⁸ This study therefore aimed to firstly translate the original GDMKQ to Afrikaans and isiXhosa and cross-culturally adapt the English, Afrikaans and isiXhosa versions for a South African population. Secondly, it aimed to preliminarily test the internal consistency and test–retest reliability of the English, Afrikaans and isiXhosa versions of the South African GDMKQ (SA-GDMKQ). The simultaneous translation of the GDMKQ into Afrikaans and isiXhosa languages and the adaptation of the English, Afrikaans and isiXhosa versions of the SA-GDMKQ are based on the fact that South Africa has 11 official languages, of which Afrikaans, English and isiXhosa are the most prevalent languages spoken in the western parts of the country.¹⁹

A prospective, observational study was conducted between February 2019 and October 2019, at a high-risk antenatal clinic situated in the Western Cape province of South Africa. The high-risk antenatal clinic is situated at South Africa’s second-largest public teaching hospital, which mainly serves patients from lower-income households from the surrounding and broader Western Cape province areas.²⁰ Many of the patients who attend the clinic are from rural and/or remote areas within the catchment area of the hospital. An initial total sample size of 86 participants was calculated based on the intraclass correlation coefficient (ICC), assuming a significance level of 0.05 and power of 70%.²¹

Participants were consecutively recruited and stratified into three language groups, namely, English, Afrikaans and isiXhosa. The inclusion criteria were as follows: pregnant participants clinically diagnosed with GDM according to the National Institute for Health and Care Excellence criteria (fasting glucose above 5.5 mmol/L and OGTT 2-h glucose over 7.8 mmol/L)²²; participants who were able to read at a basic level in either English, Afrikaans or isiXhosa; participants who were > 18 years; and participants who attended the high-risk antenatal clinic during the study period. Female patients attending the clinic who were unable to comprehend what was expected of them if they participated in the study were excluded from the study. The study was conducted during the operating hours of the clinic, and there was no interference with the daily operations of the clinic. On enrolment into the study, participants were provided with an informed consent form and a socio-demographic questionnaire aimed at extracting the following data: age, education level, living area, number of pregnancies, duration of pregnancy, language, ethnicity, employment status, among others. These documents were provided in the preferred language of the participant. The study consisted of three phases. Phase 1 involved the translation and cross-cultural adaptation of the GDMKQ, phase 2 involved the preliminary testing of the newly adapted questionnaires’ face and content validation, and phase 3 involved the evaluation of psychometric properties of the newly adapted questionnaires in terms of internal consistency and test–retest validation. Figure 1 illustrates the study phases and study flow.

This study found that all three versions had a good face and content validity, were applicable for the South African context, were easy to understand and were deemed feasible to assess the knowledge of pregnant women with GDM in the Western Cape of South Africa, although internal consistency and test–retest reliability require further investigation in larger sample sizes.

South Africa is a uniquely diverse country, consisting of multiple religions, cultures and traditions. Simultaneous translations in this study are comparable to previous South African research studies where multiple translations and simultaneous cross-cultural adaptation were conducted.³³ This said, however, translation and cross-cultural adaptation are definitely more complex in a context like South Africa.^18,34,35 The significant diversity, multiple national languages spoken and existence of various cultures and beliefs make it hard to conduct studies and to be inclusive. Typically, this complexity becomes a critical aspect of any research conducted in South Africa. With multiple translations and multiple cross-cultural adaptations, the cost of the research becomes significant, the sample size is influenced, and the research processes and outcomes are often affected. In addition, the processes are often compared to Western world research and not considered within the context. Many questionnaires and measures are developed in the less diverse Western world and in one language,^18,34,35 and very little guidance exists regarding such translation, especially for the adaptation of a questionnaire from one English population to another.¹⁸ The processes followed in this study can therefore help inform logistics for larger future studies conducted in similar contexts and reduce initial layout costs. By considering the complexities in translating and adapting tools for such a diverse population, research in other diverse countries and contexts where situations can be challenging because of similar reasons can be informed.

The SA-GDMKQ demonstrated good face and content validity. With regard to face validity, all the participants felt that the SA-GDMKQ was clear and easy to understand and that all three versions were structured and designed in a way that made them easy to complete. In a context where a majority of the population are not well educated, access to healthcare is limited, capacity and time are limited for healthcare workers, etc., the ease of using a questionnaire is vital.^18,34 The adapted versions of the SA-GDMKQ were not time-consuming and took a mean (s.d.) of 7.53 (1.36) min to complete. If the healthcare worker needs to assess a patient’s knowledge of GDM, they would be able to do so without taking too much time off their other daily tasks or from the patient’s consultation time. There was good agreement between the participant subgroups (per language) regarding the importance of including the various questions in the adapted versions of the SA-GDMKQ. The content validity of the three adapted versions of the SA-GDMKQ was therefore deemed appropriate. The three versions of the SA-GDMKQ could be recommended as practical questionnaires, especially in South Africa, where there are limited resources and staff, as well as capacity and time. Although further research is required, the results of the study provide preliminary evidence for the usefulness of such a questionnaire within the South African context.

The final versions of the SA-GDMKQ demonstrated low to moderate internal consistency; α = 0.534 for the Afrikaans version, α = 0.434 for the English version and α = 0.621 for the isiXhosa version (n = 94 participants). The internal consistency of the original GDMKQ was reported as α = 0.77, which is higher than reported in this study.¹⁷ According to our knowledge, the GDMKQ has not previously been translated or adapted, and this was the first attempt. The lower internal consistency values in this study could be because of the low numbers per language group and the low items as a whole for the questionnaire.²⁷ Larger sample sizes in consequent studies are therefore encouraged to confirm or negate the findings in this study. However, this said, the lower internal consistency values should still be considered in the context of the type of questionnaire and the type of questions asked. Some questions were on symptoms, others were on nutrition and others related to the condition itself. For this reason, a lower internal consistency between questions could be expected because a respondent might have known more about one aspect of the condition than the other. In the current context, psychometric measures such as internal consistency should be viewed with caution, and all measures should be considered as a whole rather than be an absolute determinant for reliability testing.

All versions of the SA-GDMKQ demonstrated a statistically significant positive linear correlation between the total scores obtained in the first attempt, compared to the total scores obtained in the second attempt (p < 0.001). The correlation of the English version was (rho = 0.67, p < 0.001), the Afrikaans version was (rho = 0.73, p < 0.001) and the isiXhosa version was (rho = 0.77, p < 0.001). The Spearman’s correlation coefficient values, however, do not provide any insight into systematic errors that may be inherent in the measurement obtained during the completion of the SA-GDMKQ.²⁸ Nevertheless, it did provide insights into the test–retest reliability of the adapted versions. Fluctuations of kappa scores of the SA-GDMKQ were seen between the different versions as well as between questions. This implies that the participants guessed the answers, even though the instructions to the participants were not to guess and choose the ‘I don’t know’ option. Furthermore, there were also kappa values less than 0. This raises concerns as it means that there is no agreement between the first and second attempts, indicating lower test–retest reliability. The kappa statistic considers the possibility of guessing, but the assumptions about the independence of the different attempts and other factors are not well supported, which lowers the estimate of agreement.²⁸ Our interpretation of these findings may be that there could have been external or internal factors that influenced the participants’ responses on each of the two occasions. For example, participants did not receive their answers in the first round; therefore, they would not have known what they had correct. On the second attempt, they may have thought to choose another answer, in case the first one was incorrect, or they may have thought about the question during the 2-week period and either logically thought about it, asked their fellow patients, their families or the physicians if time allowed or may even have ‘Googled’ the answers, even though they were told not to. We, however, have no way of ascertaining if this was the case, and thus, for the purposes of the study, the methods of testing test–retest reliability were deemed adequate, although the results should be read with caution. Future studies should consider the possibilities of specific factors influencing test–retest reliability of similar questionnaires and how to mitigate against such influences.

This study had a few limitations that could be considered in future in similar research studies. Firstly, the exclusion of participants who displayed a ‘lack of comprehension’ could have possibly introduced bias. In the future, it is advised that researchers carefully consider how comprehension level is ascertained and how the exclusion of participants occurs. Secondly, it may be important to consider that different settings (such as the participant’s home or the clinic) may affect test–retest processes and results and should carefully be considered. Lastly, on preliminary examination, it was found that factor analysis was not appropriate as the sample size was inadequate. Future studies should therefore consider larger sample sizes, and factor analysis should be run to confirm the construct validity of the adapted questionnaire. However, as this study was a preliminary attempt, many of the logistical challenges associated with this type of research in such a unique population could be identified, and recommendations on how to improve upon these challenges when conducting research in similar contexts should be considered prior to broader testing. Therefore, although this study has limitations, it does provide a basis for future, larger studies on which the logistics of the research can be guided.

In conclusion, this study found that the SA-GDMKQ version has a reasonable face and content validity, is applicable, is easy to use and understand, and is deemed feasible to assess the knowledge of women with GDM in the Western Cape area of South Africa. Further validation of the psychometric properties of the English, Afrikaans and isiXhosa versions of the SA-GDMKQ among larger sample groups is, however, warranted. It is envisaged that although this study was conducted in a small part of South Africa, the processes followed and challenges faced can be extrapolated to other parts of the country, and similar contexts across the globe, to inform research and policy when adapting outcome measures and questionnaires for specific populations. It is further recommended that in practice, the focus is directed to the accurate and appropriate assessment of knowledge among those living in rural and/or remote areas as accurate self-management education programmes in this population are vital.