Health-related quality of life and associated factors among patients with diabetes mellitus at the University of Gondar referral hospital

An institution-based cross-sectional study was conducted at the University of Gondar referral hospital chronic illness follow-up outpatient clinic from April to May 2017 to assess HRQOL. The hospital, located in North Gondar zone, is one of the tertiary level health care facilities in Ethiopia. It has an outpatient department for chronic illness follow up and diabetes treatment provided 2 days a week to an average of 900 diabetic patients per month. It also has inpatient facilities where medical care is provided throughout the week.

The population was all adult diabetic patients in the chronic illness follow up clinic during the study. All adult diabetic patients on follow-up for at least 6 months were included, whereas individuals with gestational diabetes and patients who were unable to communicate were excluded. The sample size was determined using a power approach (double population formula) from a previous study [23] by considering type I error of 0.05, type II error of 0.10, confidence interval 95%, standard deviation (SD) one 17.22, SD two 13.79, mean difference 5.2 and non-response 10%. Therefore, the final sample size of the study was 416. All diabetic patients who came to the University of Gondar referral hospital for follow ups were recruited consecutively until the minimum required sample size was reached.

The World Health Organization quality of life (WHOQOL-BREF) the short version of the WHOQOL-100 SCALE was used to collect data. The questionnaire which contains 26 items was developed with 15 international field centers to obtain an assessment tool applicable cross-culturally [24]. The WHOQOL-BREF contains four specific domains (physical, psychological, social and environmental). The questionnaire was first adopted in the English language and translated to Amharic and back-translated to English to maintain its consistency. Factors relating to socio-demographics (age, sex, marital status, educational level, occupation, residence, ethnicity, religion and wealth index), clinical data (duration of diabetes, type of diabetes, diabetes-related complications, co-morbidities: any chronic diseases other than diabetes mellitus, body mass index, type of treatment and fasting blood glucose) and Lifestyle (smoking, physical exercise, diet, foot care and alcohol consumption) were included in the questionnaire.

Data on patient socio-demographics, lifestyle, HRQOL and some clinical data were collected by a trained interviewer, while some clinical data (co-morbidities, diabetes-related complications, diabetes type and fasting blood sugar) were taken from patients’ medical record cards.

The collected data were checked for completeness. Then, codes were given to each question and entered into Epi-Info Version 7 Software. Further analysis was done with Stata version 12. Where an item was missing, the mean of other items in the domain was substituted. But when more than two items were missing from the domain, the domain score was not calculated with the exception of domain 3, where the domain should be calculated if only one item is missing. Negatively framed questions (items 3, 4 and 26) were transformed into positively framed ones. Cronbach alpha was used to check the reliability of the items and domains. Raw and transformed scores were considered for the outcome variables. Summary statistics were done for the outcome and independent variables. Model assumptions (normality, equal variance, multicollinearity and linearity) were checked. Simple linear regression analysis was done to identify factors associated with each domain of HRQOL independently at a P-value < 0.2. Variables that were significant at a p-value of < 0.2 were selected for the final multiple linear regression model. In the multiple linear regression analysis, variables with P-values of < 0.05 were considered statistically significant.

A total of 416 diabetic patients participated in the study. Eight (1.92%) questionnaires were excluded from the analysis because they were incomplete. Of the participants, 54.7% were male, and 33.7% were unable to read and write. Their mean age (SD) was 47.48 ± 14.9 years (Table 1).

Approximately, 56.6% of the participants were type 2 diabetes; 28.92% had co-morbidities, and 21.57% developed a diabetes-related complication (Table 2).

In this study, 41.91% of the participants rated their quality of life as good, and 18.14% were satisfied with their current health status (Table 3).

The four domains had good internal reliability with Cronbach Alpha: physical α = 0.77, psychological α = 0.69, social α = 0.73 and environmental α = 0.71. out of the 4 domains, the environmental domain HRQOL had the lowest mean score. In contrast, the social domain of HRQOL had the highest score (Table 4).

Age had a significant association (B = −.13, 95% CI = − 2.5, − 0.1), (B = −.16, 95% CI = −.30, −.01) and (B = −.20, 95% CI = −.34, −.05) with physical, psychological and social domains, respectively. Being single (B = − 6.70, 95% CI = − 12.43, −.97) and being widowed (B = − 6.64, 95% CI = − 12.6, -.613) had a significan association with the social and psychological domains, respectively. The environmental domain was significantly associated with secondary and above education level (B = 3.87, 95% CI = .14, 7.60), residence (B = − 3.88, 95% CI = − 7.72, −.04) and foot care (B = 1.12, 95% CI = .55, 1.70). Co-morbidity and occupation had a significant associate with physical domaim (B = 5.55, 95% CI = 2.55, 8.55). Diabetes-related complication statisticaly associated (B = 4.5, 95% CI = 1.17, 7.83), (B = 7.69, 95% CI = 3.18, 12.2) and (B = 3.88, 95% CI = .99, 6.77) with physical, social and environmental domains, respectively. Exercise was associated (B = 0.89, 95% CI = .08, 1.69) and (B = 1.28, 95% CI = .28, 2.29) with the physical and psychlogical domains, respectively. General diet had also a significant association (B = 0.84, 95% CI = 07, 1.61), (B = 1.14, 95% CI = .08, 2.20) and (B = 1.41, 95% CI = .71, 2.10) with the physical, psychlogical and environmental domains, respectively. Being hazardous drinker was statistically associate (B = 8.14, 95% CI = 4.08, 12.2) with the psychologicla domain (Table 5).

This was a cross-sectional study which was able only to detect associations, but not causalities. In addition, some important variables, like lipid profile and HgA1c were not included. As the study was conducted in one setting, findings might not be representative of the diabetic patients in other settings.