2.1. Typology of medical questionnaires

Medical questionnaires differ in the degree of their generality/specificity, formality/informality, and respondent’s involvement due to an immediate appeal to his/her values, attitudes, and stereotypes reflected in the questionnaire’s items. Irrespective of the instrument’s type, its focus is the presence (treatment, management), absence (prevention), or potential recurrence of acute/chronic diseases.

The four general types of questionnaires are as follows:

Quality of life (QOL) instrument/measure is an assessment of at least five life categories, including a respondent’s biological, psychological, interpersonal, social, and economic experience. The most frequently utilized categories are Health, Self-Esteem/Wellbeing, Community/Productivity, Social/Love Relationships, and Leisure/Creativity.

Health-related quality of life (HRQOL) instrument/measure is an assessment of the extent to which the personal well-being may be affected by a disease, disability, disorder. The negative effect may be assessed in the short or longer perspective.

Patient-reported outcome (PRO) instrument/patient-reported outcome measure (PROM) refers to a questionnaire presupposing responses collected directly from the patient.

Patient-based outcome assessment (PBOA) instrument refers to a questionnaire covering issues of specific concern to the patient.

Despite their functional relatedness and similarity of conceptual focus, these types of questionnaires reveal individual features. For instance, HRQOL instruments are though multidimensional, still primarily focused on the respondent’s health status and influence of disease and impairment, while QOLs investigate the individual ability to fulfill his/her needs and emotional response to the restrictions. PBOAs cover issues of particular relevance to the patient, while PROs/PROMs imply only that the information, however relevant or trivial, is provided directly by the patient.

The content and intended purpose of the medical questionnaires found their reflection in a classification by Fitzpatrick et al. which mentions disease-specific, site or region-specific, dimension-specific, generic, individualized, utility instruments, and those with summary items (Fitzpatrick et al. 1998, 8).

The design of questionnaire items and response options encompass the commissioner’s assumptions about the scope of the receptive audience’s background knowledge and health literacy. The items may be of an open or closed (dichotomous) type. Open questionnaires encourage qualitative answers, general comments by the respondent, and for this reason are harder to analyze in statistical parameters. By contrast, closed questionnaires appear to be more structured, as they are based on “yes/no answers”, with numerical values attached to them. Dichotomous type of survey instruments is associated with a high reproducibility and reliability (Gilkison, Fenton, and Lester 1992).

Response options take into account, among such other parameters as a degree of language proficiency, expertise, background knowledge, etc. – the extent of respondent’s general literacy. They may be primarily verbal (Likert scale, recording of events), verbal–numerical [VAS scales, anchored (categorized) VAS scales], primarily numerical (rating scales), pictorial scales, and checklists.

3.1. Equivalence and translation methodology

The concept of equivalence lies at the heart of translation methodology and paradoxically creates a dividing line between the classical oppositions: adequacy-oriented translation versus acceptability-oriented translation, overt versus covert, indirect versus direct one. Pym defines equivalence as “a relation of ‘equal value’ between a start-text segment and a target-text segment” which can be established on any linguistic level, from form to function (Pym [2010] 2014, 6). He also distinguishes between “natural equivalence” existing prior to the translating act and “directional equivalence” which depends upon the choice that the translator makes among several translation solutions (Pym [2010] 2014, 24).

Analysis of the translated versions of medical questionnaires shows that some of them follow quite closely the original text, without making any attempt at linguistic and cultural adaptation, while others transplant the foreign survey tool into the receiving culture. The relatively small number of the latter is explained by a high status of the original texts considered “gold standards” in the field, and the problem of validation as directional equivalence, unlike the natural one, does not reveal itself in back translation.

According to G. Toury’s classification, the first group is made of adequacy-oriented translations, i.e., focused on the exact reproduction of a source language text’s (prototext’s) features (Toury 1995, 56). On the other hand, acceptability-oriented translations seek to meet the requirements of the target culture receiving the metatext (target language text).

House (2014) finds proof of the two former orientations in the impression the target version makes upon the receiver. Overt translations, for her, signal in a variety of ways that they are target (reproduced) texts. By contrast, the covert ones read like the original texts and do not give any indication of their “secondary”, processed character. Although Harkness and Schoua-Glusberg (1998, 105), inspired by House’s classification, say that “unless there is a valid reason why respondents should consider the origins of the questionnaire, […] survey translations should be covert”, the current practice of questionnaire translation manifests the opposite approach.

Rather, as Gutt suggests, the translators should strive toward “directness” within the “overt translation” category as it is the direct translation that refers to the original context and thus “creates a presumption of complete interpretative resemblance” (Gutt [2014] 2000: 196). The receiving audience understands that they are reading a target version of the questionnaire but feel that they get the same idea as the readers of the original survey tool and, what’s important, the idea that the authors intended to convey.

This similarity of conceptual background is created by a masterful use of equivalences. Guillemin, Bombardier, and Beaton (1993) describe semantic, grammatical, and idiomatic ones which are found in almost every type of the translated text. By contrast, the equivalences typical of the questionnaires are, according to Guillemin, Bombardier, and Beaton (1993), experiential and conceptual equivalences, both vital for the cross-cultural adaptation. Experiential equivalence requires that “the situations evoked or depicted in the source version should fit the target cultural context” (1993, 1423). If no corresponding feelings or activities are found, the items should be discarded.

Conceptual equivalence refers to “the validity of the concept explored and the events experienced by people in the target culture, since items might be equivalent in semantic meaning but not conceptually equivalent” (1993, 1424). Thus, conceptual equivalence becomes a corner-stone and integral part of linguistic validation, confirmation of the target version’s status as a rightful counterpart of the original.

3.2. Requirements of accuracy and methods of translation quality assurance

The task of evaluating the quality of medical questionnaire’s translated version appears notoriously difficult, since it is not entirely clear what the focus of evaluation should be. On the one hand, an accurate translation is to convey the original text’s meaning, on the other – a fluent output is desirable as it could be read easily.

These two goals, adequacy and equivalence (often called “fidelity” and “fluency”), are not easy to reconcile. Equivalence (in its most general sense) stipulates the degree to which the output is well formed (in compliance with the target language’s grammar, lexis, syntax, etc.). Adequacy refers to the extent to which the output communicates the information present in the reference translation.

The measure of adequacy is the equivalence between the meaning of the original questionnaire’s item and the meaning of the translated one. Thus, adequacy is the goal and result of the accurate translation, while equivalence is the means of achieving this goal.

According to Txabarriaga (2009, 3), the real indicators of proficiency in translation are knowledge of the subject matter, knowledge of relevant terminology, the ability to discern meaning in context, and transfer it within the target language constraints, i.e., accurately (all meaning has been transferred), precisely (all nuances of the language, tone, intent, style have been preserved in the target language), correctly (grammar, syntax, orthography rules have been observed), completely (no part of the original was omitted and nothing has been added to the target text), and consistently (specific terms, stylistic elements, and language-specific norms have been consistently used throughout).

The IMIA Guide on Medical Translation asserts that the ultimate test for a medical translator is the validation by the end user, i.e., whether the person in need of translated materials can adequately comprehend the information provided and act according to expected results (Txabarriaga 2009, 8). Guidelines about how to produce easily read translated versions of questionnaires comprehensible to a majority of people suggest using language which could be understood by 10–12-year children, or between the sixth and eighth-grade reading levels.

Recommendations for the translators of medical questionnaires include: using short sentences with key words in each item as simple as possible; active rather than passive voice; repeated nouns instead of pronouns; and specific rather than general terms. Translators should avoid using metaphors and colloquialisms; subjunctive mood; possessive forms; vague terms; and sentences containing two different verbs that suggest different actions (Guillemin, Bombardier, and Beaton 1993).

To assure the quality of the questionnaire’s target-language version, Swaine-Verdier et al. (2004) suggest recruiting a group of 5–7 translators with varied profiles. They are to be informed of the model underlying the questionnaire, its design and content, potential respondents, etc. The quality control is effected first by the coordinator, and once the draft version has been approved, – by the lay panel of the target-language speakers who have access to the translated questionnaire (and not the original).

The coordinator-mediated discussions of the translators and potential end-users (represented by the focus group) result in a number of adaptations introduced into the draft text. Tuleja et al. mention factual reporting-driven adaptations which refer to adjustments resulting from specific linguistic and cultural practices; language-driven adaptations which refer to adjustments resulting from linguistic characteristics such as the presence or absence of the gender category; convention-driven adaptations which refer to adjustments resulting from whether the orientation of the questionnaire changes from left-to-right to right-to-left, for example, when translating from English to Arabic, or from left-to-right to top-to-bottom when translating from English to Chinese; and culture-driven adaptations which refer to adjustments resulting from the different norms, customs, or practices of a given people (Tuleja et al. 2011, 400).

Deficiencies of translation and unsuitable item content constitute the so-called nuisance factors that, according to Van de Vijver, may affect the measured results and level of comparability of data across cultures (2003, 207). However, even when the requirement of translation’s accuracy is met, a questionnaire developed in one language/culture may not necessarily “travel well” across cultures due to differences in meaning and interpretation (Braun 2003, 137). Psychometric studies help to establish correspondence of the original and translated versions and to perform validation of the questionnaire’s reliability.

4.1. Creating a valid translated version of medical questionnaire

The process of creating a valid translated version of medical questionnaire involves several stages: (1) preparation, (2) forward translation (more than one), (3) reconciliation, (4) back translation, (5) back translation review, (6) harmonization, (7) cognitive debriefing, (8) review of cognitive debriefing results and finalization, (9) proofreading, and (10) final report (Wild et al. 2005, 96–97).

The preparation stage presupposes obtaining permission to use the medical survey instrument in a different linguistic/cultural setting and inviting the instrument developer to be involved. Those in charge of the translation project develop explanation of the key concepts, and recruit the in-country persons to be actively involved in or managing the translation. The staff must meet the following requirements: (a) to be native speakers of the target language, (b) to be fluent in the source language, (c) to reside in the target country, (d) to come from a medical/health/psychology/social science background, and have experience in translating/managing the translation of PRO measures.

The people in charge of forward translation should represent multiple countries of origin, especially the target countries, for the study. Wild et al. (2009, 437) suggest more than two forward translations if many countries need to be included.

Reconciliation resolves discrepancies between the original independent translations and seeks agreement between individual speech habits and preferences.

Back Translation provides a quality-control step demonstrating that the quality of the translation is such that the same meaning may be derived when the translation is moved back into the source language. Back translation is reviewed by several experts who provide their opinions on the best possible options and identify country- or region-specific problems of translation. When it is impossible to find any solution, different final versions can be produced which maintain most of the same wording but include the country-specific variations required (Wild et al. 2009, 437).

At the harmonization stage, back translators representing each language provide a back translation of each item. Close attention should be paid to the correspondence of each back translated item to the original version as well as to any instances or trends of differences between language versions in their rendering of the concepts.

When a new medical questionnaire is being developed for use in a future clinical trial, it is necessary for the developer to interview patients with the same condition that will be studied in the trial. This target population is viewed as an expert patient group capable of describing symptoms, health status changes, and functional change for the disease or condition in question.

As soon as the interviews are conducted and the analysis completed, a draft copy of the items, or questions, can be tested with yet another patient group to see if the ideas resonate with them. The second group of patients, who must meet the same criteria for disease severity as the original population, is asked to review the draft instrument and respond to a preset series of questions.

These interviews are referred to as cognitive debriefing because of the type of questions being used – structured questions rather than open-ended questions. The interviewer asks the participants to provide thoughts about the meaning of each item and to comment on comprehension. The goal is to determine if the participants understand the questions in the same way that the original group of patients understood and described it.

Cognitive debriefing may have two distinct purposes: (1) to develop items for a new or modified questionnaire, and (2) to ascertain equivalence during translation, perform linguistic validation, and cultural adaptation of an instrument. Each of these processes uses similar methods to interact with study participants but they are completed at different intervals in the development and use of a questionnaire.

If items are not generated in all language groups included in the clinical trials, the appropriateness of the content should be addressed in cognitive interviewing in each language group tested. An item tracking matrix may be helpful to document the changes or deletions in items and the reasons for those changes (FDA 2014, 13).

Saturation is reached at the point when no new relevant or important information emerges and collecting additional data will not likely add to the understanding of how patients perceive the concept of interest and the items in the questionnaire (FDA 2014, 12).

In large multi-country trials, in case a universal version of the medical questionnaire is decided upon, proofreading is performed by representatives of different countries. The finalized text and results of cognitive debriefing are arranged into a report and sent to the commissioner/developer.

5.1. Overview of the material

The IOF’s 1-min osteoporosis risk test is a shorter (10-item) version of the questionnaire designed to predict the risk of osteoporosis in representatives of various gender, age, ethnic, etc. groups opting for various lifestyles released by the organization in 1999¹. Osteoporosis is a principal problem of health care in the developed countries; however, as a tool developed in English, the survey’s use as a validated instrument has at first been limited to the English-language populations.

Although originally conceived as an awareness-raising tool for the lay receivers (and thus requiring no validation), the IOF’s test was proved to have a considerable predicting power if used in conjunction with densitometry (Povoroznyuk and Dzerovych 2008).

Currently, the IOF’s 1-min osteoporosis risk test is performed in 96 countries with 9 translated versions. The present test includes 10 questions (7 – general ones, 2 – intended for women, 1 – intended for men). Numerical values were attached for the sake of statistical analysis (“yes” – 2 points, “no” – 1 point) [Appendix 1].

Ukrainian version was translated and adapted by Povoroznyuk R.V., Dzerovych N.I., and Povoroznyuk V.V. It was published in 2006, and posted on the website of the Ukrainian Association of Osteoporosis (https://osteoporos.com.ua) [Appendix 2].

5.2. The stages of Ukrainian version’s validation

At the first stage of validation, the translated version is to pass an ethics review by an independent ethics committee, board or governmental body. Its aim is to evaluate the relevance of the survey text’s content, its applicability for various groups of population. The Declaration of Helsinki on human research ethics (World Medical Association 1964) asserts the protection of free will, confidentiality, privacy, and well-being of the participants, although regulations vary across the countries and organizations within the same country. Pennell et al. (2010, 285) encourage the national ethics committees to pay especial attention to the sensitive items addressed by the questionnaire as these are often country or region-specific.

In the case of the Ukrainian version, the approval was obtained through the Ethics Committee by the Institute of Gerontology (Ukrainian Academy of Medical Sciences) under whose aegis the Ukrainian Scientific-Medical Centre for the Problems of Osteoporosis operates. Signing informed consents was optional as it is not stipulated by the national regulations.

The pretesting stage involved cognitive interviews with 30 women and 26 men aged 20–79 years. Their aim was to check general understanding of two forward translations (one – made by a physician with no translation training, and another – by a translation student at the Kyiv National University, Ukraine), later submitted to a back-translation by a different expert.

As Questions 1, 2, and 4 showed minor divergences in the texts of forward and back translations, their validity was further checked by means of sensitivity and specificity analysis (the total number of participants – 830 people [Appendix 3])².

The final validation stage involved 353 women and 104 men aged 20–79 years. In the clinical setting, a translator was reading questions to the patients, and an attending physician was providing professional explanations if called upon.

The data were analyzed using “Microsoft Excel” and StatSoft Statistica v6. Survey results were further compared with densitometry test findings.

5.3. Results of the case study and their discussion

(1)Question 1 (Have either of your parents been diagnosed with osteoporosis or broken a hip after a minor bump or fall?/Чи мав хтось із Ваших родичів діагноз остеопороз чи перелом стегнової кістки після мінімального удару чи падіння?) and Question 2 (Have you broken a bone after a minor bump or fall?/Чи був у Вас перелом кісток після мінімального удару чи падіння?) showed minor semantic divergences in the texts of forward and back translations (родичі indicates a more distant degree of blood relations than “parents” which makes the question more generalized and potentially leads to obtaining a broader anamnesis).

Question 4 (Have you lost more than 3 cm in height?/Чи зменшився Ваш зріст більше ніж на 3 см?) was submitted to a grammatical transformation with a themo-rhematic change of functional perspective required by the rules of usage in Ukrainian. There were no semantic changes evident in back translation.

However, to ascertain the validity of translation, sensitivity, and specificity analysis was used. It was aimed at revealing a concordance of survey results (in terms of Questions 1, 2, 4) and densitometry test findings.

Note:

Sn – the sensitivity of IOF’s 1-min osteoporosis risk;

Sp – the specificity of IOF’s 1-min osteoporosis risk;

Q1 – Question 1 (“Have either of your parents been diagnosed with osteoporosis or broken a hip after a minor bump or fall?”);

Q2 – Question 2 (“Have you broken a bone after a minor bump or fall?”);

Q4 – Question 4 (“Have you lost more than 3 cm in height?”).

(2) It is inevitable that some terminological units are more readily accepted by certain cultures than the others. Sometimes this phenomenon is due to the general level of the population’s health literacy; more often it has cultural origins. Wild et al. indicate that “wording (of the items) may not sound as natural to patients, and the language may be culturally and linguistically bland” (Wild et al. 2009, 436). However, they warn against the changes when “patients are able to understand (the items) by using their passive vocabulary, which refers to understanding terms even if not part of the patient’s everyday usage” (Wild et al. 2009, 436).

(A) Inadequate grasp of medical terminology proved to be an obstacle not only for the respondents, but also for a translator with no medical background. The back-translated Question 3 (Have you taken corticosteroid tablets for more than 3 months?) in Ukrainian included the noun гормони (hormones) instead of “corticosteroids”. The suggestion confounded the questionnaire’s results, as corticosteroids have a direct adverse effect on the bone tissue.

For those cases when misunderstanding of terminology was admitted by the respondents, presence of the attending physician proved a beneficial factor. Difficult items were explained, and potential misrepresentations avoided (Figure 1).

Figure 1.

Ratio of positive/negative answers to Question 3 (Have you taken corticosteroid tablets for more than 3 months?/Чи приймаєте Ви кортикостероїди (кортизол, преднізолон та ін.) понад три місяці?) given by male/female (a/b) respondents.

(B) While answering Question 8, female respondents (primarily from the rural areas) were noticeably grappling with the term “menopause”. To make the item more accessible, researchers tried several tactics: (1) to explain the meaning of the term descriptively, (2) to use an obsolescent, though more accepted in the Ukrainian colloquial usage equivalent клімакс.

Herget and Alegre (2009) observe that medical terms of a Greek or Latin origin are typical of an erudite communication, and thus are used in translations made for professionals; on the other hand, if the target text is addressed to a general audience, the translator should make use of lexemes originating from his/her native tongue.

After a lengthy deliberation, nevertheless, the researchers opted for the Ukrainian equivalent of the same Latin extraction as the terminological unit in the original (Do you undergo menopause before the age of 45?/Чи настала у Вас менопауза до 45 років?) (Figure 2).

Figure 2.

Ratio of positive/negative answers to Question 8 (Do you undergo menopause before the age of 45?/Чи настала у Вас менопауза до 45 років?) given by female respondents.

(C) In translation, Question 7 (Do you suffer frequently from diarrhea (caused by problems such as celiac disease or Crohn’s disease)?/Чи страждаєте Ви на часту діарею, обумовлену такими захворюваннями, як целіакія, хвороба Крона?) manifested two conflicting trends. According to Tercedor-Sánchez and López-Rodríguez (2012), there is a group of diastratic terminological variations, which are associated with specific demographical characteristics of the respondents (in this case, their social stratum). Our results show that quite a significant number of patients weren’t familiar with a medical term “diarrhea”. Instead, the attending therapist had to use an informal word пронос, a terminoid of folk-medical nature. However, the patients diagnosed with specific conditions outlined in Question 7 had no difficulty recognizing their names (Figure 3).

Figure 3.

Ratio of positive/negative answers to Question 7 (Do you suffer frequently from diarrhea (caused by problems such as celiac disease or Crohn’s disease)?/Чи страждаєте Ви на часту діарею, обумовлену такими захворюваннями, як целіакія, хвороба Крона?) given by male/female (a/b) respondents.

(D) A specific case of terminological occurrence in translation (though not in the original) is exemplified by Question 9 (Have your periods stopped for 12 months or more (other than because of pregnancy or menopause)?/Чи були у Вас періоди аменореї (відсутність менструацій) протягом 12 місяців та більше (не пов’язані з вагітністю чи менопаузою)?). The major part of the interrogative phrase “Have your periods stopped” was replaced in translation by a terminological unit періоди аменореї (periods of amenorrhea). Since “amenorrhea” is defined as “an abnormal absence of menstruation”, the researchers considered it to be more relevant. Nevertheless, resulting item (1) has a higher degree of technicality, (2) is less comprehensible to the respondents, and (3) manifests divergences in back translation. That’s why it was later annotated by means of a parenthetical definition (відсутність менструацій) (Figure 4).

Figure 4.

Ratio of positive/negative answers to Question 9 (Have your periods stopped for 12 months or more (other than because of pregnancy or menopause?/Чи були у Вас періоди аменореї (відсутність менструацій) протягом 12 місяців та більше (не пов’язані з вагітністю чи менопаузою)?) given by female respondents.

(3) Sensitive culture-specific items place a particular burden on the researchers as they influence the reliability and, thus, validity of the findings. Hudelson (2005) emphasizes the fact that patients feel stigmatized as a result of being affected by certain diseases considered disgraceful by the representatives of their culture, and more likely to conceal their diagnoses. Undue emotional or cognitive strains decrease the quality and completeness of data: respondents are unwilling to answer disease-related questions, or perceive that an interviewer wants them to react in a certain way (FDA 2014, 18). Wild et al. observe that some sensitive or dietary content may be highly culturally linked (e.g., sexual performance or items referring to alcohol) (2009, 436). Consequently, a bias occurs even when it was unintended by the translators or authors of the original instrument.

Very often the obstacles to validation of a particular questionnaire are caused by two cultures (original and receiving) belonging to different communication context rankings (Hall 1977). For example, the low-context cultures rely more on the implicit interactions than actual utterances. To add insult to injury, sensitive topics vary among cultures. However, a person’s health is shaped by cultural beliefs and experiences that influence the identification and labeling of symptoms; beliefs about causality, prognosis, and prevention; and choices among treatment options (Anderson et al. 2003, 74).

In the IOF’s 1-min osteoporosis risk test, there are two items of a sensitive culture-specific character: Questions 5 and 10. Both are gender-specific, which further complicates the matters.

(A) We have received but one positive answer to Question 5 (Do you regularly drink alcohol in excess of safe drinking limits?/Чи приймаєте Ви регулярно алкоголь у дозі, яка б перевищувала небезпечну?) from women aged 40–49, 60–79 years even after the “safe drinking limits” were defined. It might be explained by the negative cultural stereotypes associated with female drinking in Ukraine (Figure 5).

Figure 5.

Ratio of positive/negative answers to Question 5 (Do you regularly drink alcohol in excess of safe drinking limits?/Чи приймаєте Ви регулярно алкоголь у дозі, яка б перевищувала небезпечну?) given by male/female (a/b) respondents.

(B) Very few positive answers were given to Question 10 (Have you ever suffered from impotence, lack of libido, or other symptoms related to low testosterone levels?/Чи страждали Ви коли-небудь від імпотенції, зниження лібідо та інших симптомів, які пов’язані з низьким рівнем тестостерону?) from men belonging to any age groups. It might be explained by the fact that the lack of virility is associated with decline of physical and mental faculties in the Ukrainian psyche (Figure 6).

Figure 6.

Ratio of positive/negative answers to Question 10 (Have you ever suffered from impotence, lack of libido, or other symptoms related to low testosterone levels?/Чи страждали Ви коли-небудь від імпотенції, зниження лібідо та інших симптомів, які пов’язані з низьким рівнем тестостерону?) given by male respondents.