Translation, cross-cultural adaptation and validation of the Finnish Diabetes Risk Score (FINDRISC) for use in Brazilian Portuguese: questionnaire validity study

ABSTRACT BACKGROUND: The Finnish Diabetes Risk Score (FINDRISC) is a questionnaire that was developed by Finnish researchers to track the risk of diabetes. OBJECTIVE: To translate, cross-culturally adapt and validate the FINDRISC for use in Brazilian Portuguese. DESIGN AND SETTING: Questionnaire validity study conducted at a private university. METHODS: The Brazilian version of the FINDRISC was developed through the processes of translation, back-translation, committee review and pre-testing. Test-retest reliability was measured using the intraclass correlation coefficient (ICC), kappa coefficient, standard error of measurement (SEM) and minimum detectable change (MDC). Internal consistency was measured using Cronbach’s alpha. For construct validity, the total score of the FINDRISC was correlated with the Diabetes Knowledge Scale (DKN-A) and Diabetes Mellitus Risk Questionnaire (QRDM). Ceiling and floor effects were also evaluated in the present study. RESULTS: For construct validity and floor and ceiling effect measurements, a total sample of 107 participants was used. For reliability, a subsample of 51 participants out of the total sample was used. We identified adequate values for reliability (kappa ≥ 0.79 and ICC = 0.98) and internal consistency (Cronbach’s alpha = 0.84). Regarding the error inherent in the FINDRISC, we found SEM = 8.02% and MDC = 22.44%. There were significant correlations between the FINDRISC and the QRDM (r = 0.686) and DKN-A (r = -0.216). No ceiling or floor effects were found. CONCLUSION: The Brazilian version of the FINDRISC has adequate psychometric properties that are in accordance with the best international recommendations.


INTRODUCTION
Recently, the International Diabetes Federation reported that the number of individuals with type 2 diabetes mellitus (T2DM) had increased worldwide, especially in Brazil, which placed this country in fifth place among the countries with the most people with diabetes mellitus. 1 Given this reality, strategies are being sought in an attempt to easily and cost-effectively screen individuals with high potential to develop T2DM, in order to implement preventive measures against the onset of the disease. In this context, use of questionnaires has been an ally in screening for several other diseases. [2][3][4][5] A few questionnaires directed towards people with T2DM have been validated for Brazilian Portuguese, including the Diabetes Quality of Life Measure, 6 Diabetes Mellitus Knowledge and Attitude Questionnaire 5 and Diabetes Self-Care Activities. 7 With regard to the risk of developing T2DM, there is no validated questionnaire for use in Brazilian Portuguese for these purposes.
The Finnish Diabetes Risk Score (FINDRISC) is a questionnaire that was developed in Finnish and English in 2003 by Finnish researchers. It had the aims of tracking the risk of diabetes and stimulating the adoption of measures to prevent the onset of T2DM, especially for individuals who are at increased risk of the disease, but without the need for low-cost laboratory tests. 8 Some articles using the FINDRISC have already been published in Brazilian Portuguese, 9,10 even without proper translation, cross-cultural adaptation or validation of the questionnaire for this language. However, the FINDRISC has already been validated for other languages, including those spoken by the populations of Spain, 11 Greece, 12 Venezuela, 13 Colombia, 14 Hungary, 15 Germany, 16 Jordan, 17 China, 18  All the participants validated their participation by signing a free and informed consent statement. These participants were recruited from the university community in the city of São Luís (MA), Brazil, and from a community associated with this city.
Announcements regarding this study were disseminated through pamphlets and social media.

FINDRISC translation and cross-cultural adaptation
The process of translation and cross-cultural adaptation of the FINDRISC for use in Brazilian Portuguese followed the criteria of Beaton et al. 21 and was performed in five phases as described below: The respondents' comprehension of the items and responses in the FINDRISC was evaluated.

Participants
The sample size for this validation study was based on COSMIN, and a minimum of 100 individuals was recommended. 23 The eligibility criteria that we used were that the participants could be of either sex, without any diagnosis of type 1 or type 2 diabetes mellitus, aged over 24 years and under 64 years, and without cognitive deficits or any other limitations that would make it impossible for them to respond to the questionnaire.

OTHER QUESTIONNAIRES
In addition to the FINDRISC, we applied two other questionnaires that had already been adapted and validated for use in Brazilian Portuguese, to ascertain the construct validity.
These instruments were: which is composed of seven items, with a total score that can range from 0 to 27 points. The higher the score is, the higher the respondent's risk of developing T2DM is.

Statistical analysis
Reliability was assessed based on a test-retest model by measuring the kappa coefficient, intraclass correlation coefficient Ceiling and floor effects were evaluated in the present study.
By definition, these effects occurred when a number of study participants (set as over 15%) reached the minimum or maximum values of the total score of the questionnaire.

Sample characterization
The study consisted of three samples of different sizes. Initially, for the analysis on the pre-test version of the instrument, 30 participants were included. For the assessment of construct validity and measurement of the floor and ceiling effects, a total sample consisting of 107 participants who were evaluated at a single time was used. For determine the reliability of the FINDRISC, a subsample was used, comprising 51 participants from the total sample who were evaluated at two times (test and retest), seven days apart. The sample size used in this validation study was in accordance with international practices for cross-cultural adaptation and validation of questionnaires, as guided by the study of Beaton et al. 21 and by COSMIN. 22 The participants' characteristics are described in Table 1, along with the mean scores from the QRDM and DKN-A questionnaires.

Translation and cross-cultural adaptation
The translation and back-translation processes are described in The pre-final version of the FINDRISC was applied to 30 individuals (24 women, 80%) without a diagnosis of diabetes and with Brazilian Portuguese as their mother tongue. The average age of this sample was 49 years (standard deviation, SD = 10.82), and the average FINDRISC score was 11.56 points (SD = 4.70).
There was 100% understanding of the questionnaire items, which thus defined the final version of the FINDRISC in Brazilian Portuguese (Appendices 1 and 2). Table 3 presents the reliability values item-by-item and the total score of the final version of the FINDRISC in Brazilian Portuguese.

Reliability and internal consistency
We found adequate values for reliability (kappa ≥ 0.66 and ICC = 0.98) and internal consistency (Cronbach's alpha ≥ 0.82).
In addition, in calculating the error inherent to the FINDRISC total score, we found that the values were sufficient, as fol-

Construct validity
The construct validity was tested by correlating the total FINDRISC score with the scores from the other questionnaires   [28][29][30] Some limitations of this study and suggestions need to be noted.
Firstly, we recommend that the cross-cultural adaptation of the FINDRISC to other languages should be tested based on COSMIN. 22 Lack of such testing greatly limited the discussion of the results.
In addition, because of the need for specific methodology, our study did not investigate the accuracy or responsiveness of the FINDRISC. Thus, we suggest that future studies should measure these psychometric properties in Brazilian Portuguese.

CONCLUSION
The results from this study demonstrate that the Brazilian  Table 3. Finnish Diabetes Risk Score (FINDRISC) reliability and internal consistency with presentation of mean values, standard deviation (SD), kappa or intraclass correlation coefficient (ICC) with 95% confidence interval (CI) and Cronbach's alpha