Diabetes-Specific Questionnaires Validated in Brazilian Portuguese: A Systematic Review

ABSTRACT Two researchers conducted independent searches on five different electronic databases: PubMed/MEDLINE, Embase, SciELO, LiLACS and Web of Science. Studies were selected that covered cross-cultural adaptation methodology and validation in Brazil with type 1 and type 2 diabetes patients of any age. After reading the full-text articles, data related to psychometric characteristics were extracted from each study selected. Reliability was assessed with Cronbach’s α (Cα). The initial searches identified 2,211 studies. After exclusions, 26 were included, covering a total of 31 questionnaires. Questionnaires were grouped into 11 domains based on their main focus of interest: adherence (n = 8), quality of life (n = 7), diabetes knowledge (n = 3), hypoglycemia (n = 3), self-efficacy (n = 3), satisfaction with pharmaceutical services (n = 1), emotional stress (n = 2), hope (n = 1), attitude towards diabetes (n = 1), perception of disease severity (n=1), and risk of developing diabetes (n = 1). This study identified and reviewed all of the diabetes-specific questionnaires that have been validated for Brazilian Portuguese, which should facilitate selection of the most appropriate instrument for each domain of interest in future research and clinical settings.


INTRODUCTION
R ecent data show that the global prevalence of diabetes increased from 108 million in 1980 to 425 million in 2017 (1,2). It is estimated that by 2045, 629 million people will be affected with diabetes worldwide (2). Brazil is the country with the fourth highest number of diabetes cases in the world, with a prevalence of 12.5 million people with diabetes in 2017 (2), which has been increasing significantly over the last 35 years (3).
The special and rigorous care that is necessary for good glycemic control affects several spheres of patients' lives, such as eating, physical activity, quality of life, and mental health, among others. This characteristic of the disease requires frequent monitoring by health professionals, preferably with a multidisciplinary approach (4). Additionally, diabetes is a disease with great potential to disable, due to its complications (5), which means that frequent evaluations of every aspect of the disease are needed (4). Questionnaires, scales, and other instruments have proved to be important tools for evaluating many chronic diseases, including diabetes, in both clinical practice and research settings (6). Questionnaires are great tools for collecting information about people's behavior, knowledge and attitudes, by administering standardized questions (7).
Questionnaires can be classified as general questionnaires, which evaluate health variables and allow comparison between patients and healthy persons, or specific questionnaires, which evaluate characteristics related to a specific disease (8). Diabetes-specific instruments are very well accepted by patients and enable evaluation of specific aspects of the disease (9). These tools have been identified as one of the best ways to evaluate certain characteristics of diabetes, such as diabetes burden and mental health (10).
The number of questionnaires cross-culturally adapted and validated for the Brazilian culture is still insufficient for the high diabetes demands in Brazil (10). There are no studies comparing all the available instruments, which could facilitate selection of the most appropriate instrument for each domain of interest in future research and in the clinical setting. Therefore, the present study aims to identify and compare all diabetes-specific questionnaires validated in Brazilian Portuguese.

SUBJECTS AND METHODS
This systematic review protocol was registered on the International Prospective Register of Systematic Reviews (PROSPERO) under registration number CRD42017073407. This protocol is fully available on the National Institute of Health Research -Health Technology Assessment Database (NIHR HTA) website. This systematic review follows the recommendations of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) (11).

Information sources
The literature search was conducted independently by two reviewers on five electronic databases (Medical Literature Analysis and Retrieval System Online/ PubMed, EMBASE, Web of Science, Scientific Electronic Library Online [SciELO], and Latin American and Caribbean Center on Health Sciences Information [LILACS]) in order to identify diabetesrelated questionnaires, scales, and other instruments that have been translated, adapted, and validated for the Brazilian culture. Other mechanisms such as Google Scholar were also used in order to reduce publication bias and include unpublished data or publications not indexed in the databases listed above. Manual searches were also conducted in the references of the articles selected and in the annals of major Brazilian and international conferences related to diabetes held during the last three years. Where studies were identified, but the full text was unavailable, the authors were contacted in order to request the full article. The most recent search was performed on October 3 rd , 2017.

Search strategy
Descriptors "Diabetes", "Questionnaires", and "Brazil" and related terms and keywords were used in the search strategy, in English and Portuguese. The search strategy run on PubMed is described in Table 1.

Eligibility criteria
Studies were selected based on the following criteria: translation to Brazilian Portuguese, cross-cultural adaptation, or validation of questionnaires, scales, and other instruments related to diabetes. Only patientreported outcome measures were included. A decision was taken to exclude articles in which the instrument validated required additional information, such as the results of laboratory procedures, imaging examinations, or careful professional evaluation, thereby ruling out inclusion of risk scores. No date or language restrictions were applied.

Study selection
Duplicates were identified and excluded. Two researchers (L.G.B. and M.M.M.) independently evaluated titles and abstracts against the inclusion and exclusion criteria. Publications for which the information needed could not be extracted from the title or abstract were selected for full article reading. All studies selected in the first evaluation were referred for analysis of the full text. Disagreements were resolved by arriving at a consensus in discussions or by a third researcher (G.H.T). Agreement between the two researchers was analyzed using the kappa coefficient and a value > 0.80 was defined as almost perfect agreement (12). EndNote version X7 software (Thomson Reuters, New York, NY) was used to organize the studies.

Data extraction
Two researchers (L.G.B. and M.M.M.) separately extracted the following data from each study: title in Portuguese and English, publication date, authors, instrument name in English and Portuguese, population used for validation, target age, cross-cultural adaptation process, reliability data, type of validation, instrument application time, mode of administration (self-report or administered by an interviewer), instrument accessibility, number of items and domains, scores and cut-off points. Disagreements were resolved by discussion and, if unresolved on this basis, a third researcher (G.H.T.) was consulted. When two studies validated the same questionnaire in a similar population but with different sample sizes, only the most accurate was included. Data were inputted to a pretested Microsoft Office Excel TM spreadsheet.

Risk of bias in studies
The quality of studies was evaluated using the Cochrane tool (13). Certain adaptations were made in order to use it for validation studies, because it was originally created for evaluating randomized clinical trials. The bias risk assessment took into account presence of important methodological elements of the cross-cultural adaptation/validation process: validation additional to the cross-cultural adaptation process (14), reliability data, test-retest evaluation, pre-test evaluation, and sample size calculation or estimation.
Reliability was measured with Cronbach's α. In general, the minimum acceptable value for the reliability of a questionnaire is 0.70; below this value, the internal consistency of the scale is considered low. In contrast, the expected maximum value is 0.90; above this value, one might consider redundancy or duplication, which means that several items would be measuring exactly the same element of a construct; therefore, redundant items should have been eliminated. Usually, Cronbach's α values between 0.80 and 0.90 are preferred (8,15). We used the value 0.7 to consider a questionnaire acceptably reliable.

RESULTS
The initial searches identified 2,211 articles; 128 were duplicates and were excluded; 2,083 articles were analyzed by title and abstract, and 43 were selected for full-text reading. Two papers were found only in abstract form. Since their validation studies were not found, the authors were contacted and requested to provide them. Both (16,17) had used unvalidated instruments. In both cases, the instruments were excluded because it was impossible to analyze the studies. Twenty-four studies were included initially, and another two studies were found by manual searching, totalling a final sample of 26 studies included ( Figure 1). Although not fully published, three important questionnaires related to hypoglycemia were included since they presented sufficient information and psychometric data in the abstract (18).
The total number of instruments adapted/ validated was 31, because some articles used more than one questionnaire. Some of these validations were conducted for the same questionnaire, but in different populations, such as type 1 and type 2 diabetes [the Self-Care Inventory-revised questionnaire (19,20) and the Diabetes Quality of Life Measure (21,22)] or different age ranges (19,23), totaling 28 different instruments.
With regard to the populations used to validate the instruments, one study stated that the questionnaire had been validated for patients with both types of diabetes (type 1 and type 2 diabetes), but the sample included only one patient with type 2 diabetes (24). In this case, we included this study in the systematic review as exclusively validated for patients with type 1 diabetes. An analysis of concordance between the researchers was conducted, generating a Kappa coefficient of 0.848 (P < 0.001).
In order to facilitate understanding and enable comparisons between the instruments selected, the researchers grouped the studies into domains based on the main subject addressed. This resulted in 11 large domains representing the primary focus of each questionnaire: treatment adherence (n = 8, Cronbach's α = 0.71-0.79 for type 1 diabetes mellitus; Cronbach's α = 0.63-0.84 for type 2 diabetes mellitus); quality of life  Analyses of the risk of bias within studies found that most studies presented a low risk of bias in the methodological steps and psychometric data evaluated ( Figure 2 and Table 3). However, most of the studies lacked a methodological basis for determining sample size. Sample size was calculated in only five of the 26 studies (19,38,(41)(42)(43), which may have modified the psychometric data and constitutes an increased risk of bias. Also, during the data extraction phase, there were some difficulties in extracting certain information due to lack of clarity in the descriptions of the methodological processes.

DISCUSSION
Due to the increasing number of diabetes cases in Brazil, it has been necessary to develop specific instruments to assist health professionals to follow-up these patients. The present study analyzed diabetesspecific questionnaires available and validated in Brazilian Portuguese with the purpose of facilitating the process of choosing the best surveys to use in clinical settings and research. A total of 26 studies were found that validated one or more instruments, totalling 31 different questionnaires validated in different populations (children/adolescents or adults with type 1 diabetes or type 2 diabetes).

Diabetes-specific surveys in Brazil
Arch Endocrinol Metab. 2020;64/2   There is a discussion in the literature about using disease-specific questionnaires rather than general questionnaires (8,44). It is agreed that general instruments are useful for the comparison between individuals with and without disease. However, concern has been expressed that a generic health measure would not be sensitive enough to capture all aspects of patient experience, and that a diabetes-specific measure would be needed to evaluate the overall impact of the disease and interventions. More specific instruments have the advantage of better assessing the unique aspects of a particular disease as well as its impact on patients' lives (15), such as questions asking patients with diabetes about the burden of using insulin in public.
Adherence to treatment was the area with the greatest number of questionnaires. This is defined as

Diabetes-specific surveys in Brazil
Arch Endocrinol Metab. 2020;64/2 behavior of accepting and following recommendations (such as using medication correctly, sticking to the correct diet, practicing physical activity, etc.) provided by a health care professional (45). The importance of measuring adherence is to assess whether and to what extent recommendations are followed by the patient. Of the questionnaires covering this subject, the Measure of Adherence to Oral Antidiabetic Treatments (MAT ADO) has the best reliability. However, this questionnaire only evaluates adherence to oral medications and takes a long time to administer (61 minutes) (26). For a global evaluation of adherence, including items such as physical exercise, blood glucose testing, insulin use, and diet, the Diabetes Self-Management Profile seems to be better for assessing adherence in children and adults with type 1 diabetes (19,23). The Self-Care Inventoryrevised (SCI-R) is a widely-used treatment adherence questionnaire and may be the best instrument to assess a population consisting of both type 1 and type 2 diabetes patients (19,20).
Regarding quality of life assessment, a considerable number of questionnaires assess global quality of life or quality of life in relation to a specific complication, such as diabetic neuropathy (38). Health-related quality of life is a broad and subjective concept involving physical, psychological, and social factors, as well as the individual's perception of their role in society (46). It is important, since treatment should not aim merely to control a particular element of a disease (such as control of glycated hemoglobin and avoiding complications related to diabetes), but should also target aspects related to improvement in quality of life and wellbeing (47). The Diabetes Quality of Life (DQOL) questionnaire is a comprehensive questionnaire that assesses satisfaction with treatment care and the impact of diabetes on various areas of life such as social life, family, physical activity, sexual life, etc. There are 4 versions validated for Brazilian Portuguese, including both type 1 and type 2 diabetes patients, as well as a specific evaluation for young people, all with adequate reliability and validation items (21,22,35,36).
Another essential subject to be measured in patients with diabetes is their knowledge about their disease, since it is directly associated with adherence to treatment (48). The Diabetes Knowledge Scale (DKN-A) questionnaire was one of the first diabetesspecific questionnaires to be adapted for the Brazilian population and remains a tool of great utility and excellent reliability (28).
Hypoglycemia plays an important role in management of patients with diabetes, especially those with type 1 diabetes undergoing intensive treatment (49). Questionnaires to measure hypoglycemia objectively have been available in other countries for decades; in Brazil, three tools have recently been validated and will probably be of great utility for our population (18).
The remaining concepts measured are also of great importance in the research and clinical follow-up of patients with diabetes: attitude towards diabetes, emotional distress, hope, perception of disease severity. However, some common barriers faced by patients with diabetes, such as beliefs, relationship with health care professionals, lifestyle changes (50), physical activity (51), and risk of driving (52) have not been evaluated in Brazilian diabetic populations and still require crosscultural adaptation and validation in order to be ready for using.
This study has some limitations. First, it is limited to the evaluation of specific questionnaires in the population with diabetes, which may limit comparisons to other populations. Also, when analyzing the time taken to administer the instruments, it appears likely that some studies reported the total duration of the interviews conducted for validation of the instrument (including informed consent procedures, sociodemographic data collection, etc.), rather than only the time needed to administer the instrument itself. Because of this, we believe that the administration times for some of the instruments shown in table 1 may be overestimated. Additionally, some questionnaires were not fully validated, but translated and culturally adapted using only a pre-test evaluation, which left them with no available reliability information. Nevertheless, we believe this study extends the literature by identifying and bringing together all of the questionnaires that have been validated in Brazilian Portuguese.
In conclusion, this systematic review identified and analyzed all the diabetes-specific questionnaires that have been validated in Brazilian Portuguese in order to assist health professionals in follow-up and treatment of patients with diabetes, in both clinical and research settings.
Author contributions: all authors participated significantly in the concept and design. Analysis of data: LGB, MMM and GHT. Writing the manuscript: LGB, MMM, GHT and SPG. Editing and revising the manuscript: SPG, BDS and GHT. All authors have read and approved the current version of the manuscript.