Sampling

Purposive sampling was employed to obtain a heterogeneous sample of T2DM patients. Patients were recruited from the endocrine wards Raja Isteri Pengiran Anak Saleha (RIPAS) Hospital, the specialist diabetic outpatient clinic in RIPAS Hospital, Sengkurong Primary Health Centre and Rimba Primary Health. RIPAS is the national hospital of Brunei, which has more than 600 beds and offers a wide range of health services to the indigenous population.

The inclusion criteria were a diagnosis of T2DM, age above 20 years, and ability to read and communicate in Brunei spoken Malay. The exclusion criteria included inability to answer questions due to debilitating health conditions and the main reason for admission or visit not being T2DM or related health problems.

Data collection procedure

After informed consent was obtained, all patients were asked to complete the EQ-5D-5L and questions assessing their demographic and diabetic status in the ward or clinic where they were recruited. Assistance from a trained research assistant was provided whenever necessary. After the questionnaire survey, a saliva sample was taken from each patient and subsequently analysed using a commercial test kit for the concentration of cortisol, a biomarker for the level of stress [9].

All patients were invited to complete the EQ-5D-5L again one week after the baseline survey via telephone, post, email, test messaging or Whatsapp communication, depending on patients’ own preferences. Up to three reminders were sent to non-respondents.

Instrument

The EQ-5D-5L consists of five Likert-type items and a visual analogue scale (aka EQ-VAS). Each item asks respondents to rate their HRQol in a different dimension (mobility, self-care, usual activities, pain/discomfort, or anxiety/depression) as having no problems, slight problems, moderate problems, severe problems, or extreme problems. Responses to the 5 items can be used to calculate a single index score anchored by 0 (dead) and 1 (full health). This score indicates the utility or value of the respondent’s health from the perspective of the general public [10]. The EQ-VAS is a 20cm, vertical scale numbered from 0 to 100 for respondents to assess their overall health on the day of the survey, where 0 and 100 are labelled with ‘the worst imaginable health state’ and ‘the best imaginable health state’, respectively.

The Brunei Malay version of the EQ-5D-5L questionnaire was developed using the cultural adaptation procedure recommended by the EuroQol Group. It involved firstly proposing the wording for a Brunei Malay EQ-5D-5L based on the Singapore and Malaysia Malay versions of the questionnaire [11] by one of the investigators of this study and researchers from the Malay Language and Linguistics Department of University Brunei Darussalam who had experience in questionnaire translation. Secondly, an independent reviewer with a health background checked the Brunei Malay version for the appropriateness for use in the clinical setting. Thirdly, a pilot study was conducted with 16 Brunei Malay-speaking persons whose professions were not health related. The feedback collected from the pilot study was used to finetune the wording of the questionnaire. Lastly, the entire process was documented in a report which was subsequently reviewed and approved by a translation consultant appointed by the EuroQol Group. This was to ensure that the Brunei Malay version is semantically equivalent to the original English version of the questionnaire.

Data analysis

The study sample was described in terms of frequency and proportion for categorical variables and mean and standard deviation for continuous variables. The EQ-5D-5L health index score was calculated by first mapping the EQ-5D-5L health profiles to EQ-5D-3L profiles using an algorithm developed by van Hout et al [12] and then applying the EQ-5D-3L value set of Singapore [13], which is a neighbouring country of Brunei.

Known-groups validity of the EQ-5D-5L health index and EQ-VAS was evaluated by comparing subgroups of patients known to differ in health status. The known groups were defined according to smoking status (yes versus no) [14], body mass index (obese versus non-obese) [15], use of insulin (yes versus no) [16], duration of diabetes (<10 years versus 10 or more years) [17], presence of diabetes-related complications (yes versus no) [17], presence of comorbidity (yes versus no) [17], and level of cortisol concentration (<0.10 versus ≥0.10), which is a predictor of depression [9]. We hypothesized that both scores would be lower in patients known to have 'worse' health than those who had 'better' health. Taking insulin usage as an example, we hypothesized that patient who used insulin (the group in ‘worse’ health) would have lower health index and EQ-VAS scores than those who did not use insulin (the group in ‘better’ health’). Linear regression models were used to examine the differences in EQ-5D health index and EQ-VAS scores between each pair of known-groups with the adjustment of age, gender, and race.

Test-retest reliability was assessed for the EQ-5D-5L health index and EQ-VAS using intraclass correlation coefficient (ICC) and for the five EQ-5D-5L items using Cohen’s kappa. For both the two measures [18, 19], values being 0–0.20, 0.21–0.40, 0.41–0.60, 0.61–0.80, and 0.81–1 indicate slight agreement, fair agreement, moderate agreement, substantial agreement, and almost perfect agreement, respectively. All statistical tests were two-sided with the significance level set at 0.01.

All analyses were conducted using SAS (version 9.3).

Validity

The validity analyses were based on all the 154 patients. In all known-groups comparisons, the ‘better’ group was found to have a higher mean EQ-5D-5L index score than the ‘worse’ group (Table 3). Statistical significance was reached in 4 of the 7 comparisons (i.e. smoking status, years of DM, complications, and comorbidities). Nevertheless, the mean EQ-VAS score was higher for the ‘better’ group than for the ‘worse’ group in only 4 of the 7 comparisons (i.e. insulin, complications, comorbidities, and cortisol concentration), and none of the comparisons showed a statistical significant difference (Table 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Association of health and clinical variables with EQ-5D-5L health index and EQ-VAS (n = 154).

https://doi.org/10.1371/journal.pone.0165555.t003

Test-retest reliability

The reliability analyses were based on the 116 patients who completed both the baseline and follow-up survey. Cohen’s kappa value was 0.340 for the mobility item, 0.440 for the self-care item, 0.446 for the usual activities item, 0.206 for the pain/discomfort item, and 0.367 for the anxiety/depression item. The ICC value for the EQ-5D-5L health index and EQ-VAS was 0.626 and 0.521, respectively.