Participants
We included type 2 diabetic patients, who were at least 18 years of age, had been diagnosed for at least 12 months, resided in Quito with no intention of moving in the near future and were native Spanish speakers. Recruitment to the study used purposive sampling through a patient club for people with diabetes at the Chimbacalle Health Center and contacts from health promoters from several health centres in Quito (Número 1, Jardín del Valle, Cotocollao, Jaime Roldos Aguilera, Corazón de Jesus, Comité del Pueblo, San Antonio de Pichincha, Colinas del Norte, Pomasqui, Carcelén Bajo, El Condado, Mena del Hierro, La Bota, Pisulí, Puellaro, Chavezpamba, Cotocollao Alto and Calacalí).
In this setting, diabetic patient’s clubs are sometimes established in primary health care centres, either by initiative of the health staff or the patient’s themselves. The role of patient clubs is to motivate patients through the exchange of experiences among its members, in addition to the orientation, advice and guidance offered by health professionals on behaviour modification (physical activity/diets) [24, 25].
Our selection sought to include a group of patients that was heterogeneous in terms of sex, age and level of education. All participants gave their consent to participate in the study.
Procedure
The interviews were carried out between February and July 2020. The DHP-18 validation process consisted of 2 phases.
Linguistical and cultural adaptation
Two Ecuadorian medical researchers reviewed the original version of the DHP-18 (English) and the existing translation (Spanish for the United States) to assess the cultural and linguistic relevance for its use in Ecuador. They suggested some changes in text, as well as the reasons for these changes and provided a new recommended translation. Changes were discussed with the other members of the team and a new adapted version of the questionnaire was proposed. Subsequently, 2 different researchers carried out interviews to assess the linguistic and cultural understanding of the adapted questionnaire with 8 people with T2DM of Ecuadorian nationality in the Chimbacalle Health Centre. Participants were asked to answer the questions and then, the necessary time was recorded, the answer options were discussed, the wording that was difficult to understand was commented, and alternative wording was suggested based on the participants’ own words. A second adapted version was proposed. The interviews were recorded and transcribed verbatim for analysis. Finally, participants' responses were summarized in a pilot test report including recommended changes and suggestions. The report was then sent to the original authors of the questionnaire for verification and approval.
Psychometric validation
Firstly, we recruited 146 participants for the baseline test where they responded to the questions posed in the tool previously linguistically validated DHP-18 instrument in Ecuador and in another tool (SF-12v2 in its version for use in Ecuador) [26] in order to assess the correlation with generic quality of life as a construct validity test. Two weeks later, we assessed the intra-observer reliability of the new tool in a random sample of 75 of the previously interviewed patients, where only DHP-18 was retested, along with the following question: “Compared to the last time you completed the questionnaire, how do you assess your condition today? (1) unchanged, (2) improved, (3) greatly improved, (4) impaired or (5) highly impaired”.
Data collection
The 8 interviews carried out during the linguistic and cultural adaptation were held face to face but given the situation generated by the COVID19 pandemic [27], the data for the psychometric validation was collected through individual telephone interviews. Responses were digitally recorded by the interviewer using the Kobo toolbox (http://www.kobotoolbox.org/) free open-source software on electronic tablets. Informed consents were provided orally and were audio recorded.
DHP-18 questionnaire
Participants responded to the adapted version of DHP-18. We used the Diabetic Health Profile (DHP) -18 because it is a shortened version of DHP-1, a specific instrument for measuring the psychological and behavioural impact of type 1 diabetes. We decided to use the short version of the DHP because it can be used in people with both type 1 and type 2 diabetes aged 11 and older. And because the instrument has demonstrated adequate metric properties and its completion time is approximately 5–6 min. Items are scored using a 4-point Likert-type scale ranging from 0 to 3. Items are provided with one of four sets of responses (1) never, sometimes, generally, always; (2) never, sometimes, often, very often; (3) not at all, a little, a lot, very much; and (4) very likely, quite likely, unlikely, not at all likely. The raw subscale scores are transformed into a common score range from 0 to 100, with 0 representing no dysfunction.
The DHP-18 consists of three dimensions: psychological distress (includes questions like depressed from diabetes; more arguments or upsets at home than there would be if you did not have diabetes; losing your temper over unimportant things; etc.), barriers to activity (includes questions like food controls life; difficult staying out late; avoid going out when sugar is low; etc.) and disinhibited eating (includes questions hard to say no to food you like; ease of stopping when you eat; wish there were not so many nice things to eat; etc.).
SF-12 v2
The SF-12 v2 is an instrument for measuring health-related quality of life [26], based on SF-36. It includes twelve items, has an application time of approximately two minutes, and is used to evaluate the degree of well-being and functional capacity of people over 14 years of age. The response options form Likert-type scales (where the number of options varies from three to six points, depending on the item), which assess intensity and / or frequency of people's health status. The score ranges from 0 to 100, where the higher score implies a better health-related quality of life. The SF-12v2 has demonstrated adequate validity and reliability in the United States and internationally, and the Spanish version has been used successfully in Latin America and with Spanish-speaking populations in the United States. Investigations that use these twelve items of the SF have verified that the instrument is a valid and reliable measure in Latin American countries such as Colombia and Chile in adult population, and a translated version is available for Ecuador.
The SF12v2 includes questions related to health status and limitations in doing activities, problems with work or other regular daily activities due to physical health, due to emotional problems, pain, feelings, etc.
Sociodemographic and clinical variables
We collected sociodemographic and clinical variables (all self-reported by the participants): age, sex, marital status, ethnicity (mestizo or other minorities. The mestizos are an ethnicity composed of Spanish and indigenous heritages), educational level, monthly income, employment status, smoking status, alcohol intake, weight, height, duration of illness, use of medications, diabetes complications and comorbidities.
Statistical analysis
We included descriptive statistics through frequencies, the mean (standard deviation) or the median (interquartile range), as appropriate. The psychometric characteristics of the DHP-18 were assessed according to consensus-based standards for the selection of health status measurement instruments (COSMIN) guidelines [28]. Missing values for the DHP-18 and SF-12 v2 were substituted with the mean of the completed questions for those dimensions in which ≥ 50% of questions had been completed [29, 30].
We evaluated floor and ceiling effects by calculating the percentage of patients scoring either the lowest or highest possible dimensional scores. If more than 15% of respondents achieve the lowest or highest possible score, then floor or ceiling effects are present [31].
Statistical analyses were performed using Stata Version 15 (StataCorp LP; College Station, TX) and R software, version R 4.0.0 (R Core Team. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria; http://www.R-project.org) was used to perform the confirmatory factor analysis. The level of statistical significance was set at p < 0.05.
Structural validity
We performed a confirmatory factor analysis (CFA) because the factor structure had already been determined [32] and confirmed for other language translations [23]. In this case, we used CFA using Diagonally Weighted Least Squares (DWLS) [33,34,35] to test the hypothesis that the general construct of DHP is composed of three individual and correlated factors: psychological distress (6 items), activity barriers (7 items) and disinhibited eating (5 items). To estimate the model fit, we used the following criteria: Values > 0.95 for the Tucker-Lewis index (TLI) or for the comparative fit index (CFI) and the root mean square error of approximation (RMSEA) < 0.06 or the standardized root mean square residual (SRMR) < 0.08 are considered as a good model fit [36, 37]. The magnitudes of factor loadings of 0.3 or greater were considered suitable.
Reliability
To measure internal consistency reliability, we used Cronbach’s alpha coefficient, where values > 0.7 are considered as acceptable [36]. The homogeneity of items was verified by the analysis of item-rest and inter-item correlations for the items constituting each dimension of the scale. The usual rule of thumb is that an item should correlate between 0.3 and 0.7 with the total score of the factor (excluding that item), using Pearson’s coefficient. Additionally, average inter-item correlations for items in the same factor should correlate moderately, between 0.15 and 0.5, to ensure that they measure the same construct but not so closely as to be too redundant [38].
We measured test–retest reliability in patients reporting no-change in the global assessment of change question. To measure test–retest reliability we considered that the individual’s health was significantly better if they responded, “much better” or “somewhat better” in the global assessment, or significantly worse if they responded “somewhat worse” or “much worse” [39]. We used the intraclass correlation coefficient (ICC) under a 2-way random effects model with absolute agreement [40], and its associated 95% confidence interval. We considered that a questionnaire exhibits substantial reliability when ICC is between 0.40 and 0.75, and greater than 0.90 represents excellent reliability [36].
Measurement errors were determined by calculating the standard error of measurement (SEM) and the smallest detectable change (SDC). We calculated SEM by the square root of the error variance derived from analysis of variance (ANOVA), two-way ANOVA with repeated measures [41]. The SDCindividual and SDCgroup was calculated with the following formulas (41):
$${\text{SDC}}_{{{\text{individual}}}} = { 1}.{96 }* \, \surd {2 }*{\text{ SEM}}$$
SDCgroup = (SDCindividual /√n); n: number of subjects in the sample.
We estimated the minimally important difference (MID) for each DHP-18 dimension using three distribution-based methods to estimate MID: 0.2 and 0.5 standard deviation (SD) and SEM estimations. Formulas:
$$\begin{aligned} & 0.{\text{2SD}} = \, 0.{2}*{\text{ SD}}_{{{\text{basaline}}}} \\ & 0.{\text{5SD}} = \, 0.{5}*{\text{ SD}}_{{{\text{basaline}}}} \\ & {\text{1SEM}} = {\text{SEM}} \\ \end{aligned}$$
We also estimated Cohen’s d effect size (ES) of the change in DHP-18 dimensions for those reporting a small but important change and those reporting no changes in global assessment rating. Cohen’s d was calculated with the following formula (42):
$${\text{ES }} = \, \left( {{\text{Score}}_{{{\text{baseline}}}} - {\text{Score}}_{{{\text{retest}}}} } \right)/{\text{SD}}_{{{\text{basaline}}}}$$
SDbasaline: Standard deviation of baseline score.
An effect size of 0.2 was considered small, 0.5 moderate and 0.8 large [43].
Construct validity
We assessed construct validity of the DHP questionnaire using three approaches. Firstly, we assessed convergent validity using binary correlation analysis (Spearman’s r- due to non-normal value distributions) of the DHP-18 and SF-12v2. Before starting the analysis, we set up the following a priori hypothesis: (1) Scores of “psychological distress” dimension in DHP-18 correlate negatively with scores of “mental health” dimension in SF-12v2. (2) Scores of “activity barriers” dimension in DHP-18 correlates negatively with “physical dimension” in SF-12v2. (3) Scores of “disinhibited eating” dimension in DHP-18 correlates negatively with “physical dimension” in SF-12v2.
Secondly, we explored discriminant validity by comparing the correlation among the three dimensions of the DHP-18 scale.
Thirdly, we evaluated known-group validity by comparing DHP-18 scores in patients according to sex, education level, obesity, and clinical characteristics such as duration of diabetes, presence of comorbidities and/or diabetes-related complications using a Student’s t-test or ANOVA. We tested the following pre-defined hypotheses:
H1: Individuals with longer duration of illness would have higher DHP-18 scores (poorer quality of life) than those with shorter illness duration [44].
H2: Obese individuals would have higher DHP-18 values (poorer quality of life) than non-obese individuals [45].
H3: Women would report higher DHP-18 values (poorer quality of life) than men [46].
H4: Individuals with comorbidities would have higher DHP-18 values (poorer quality of life) [44].
H5: Individuals with a higher education level would have lower DHP-18 values (better quality of life) than those with a lower education level [47, 48].
H6: Individuals with diabetes-related complications would have higher DHP-18 values (poorer quality of life) than patients without complications [44].