Depression among patients with diabetes mellitus in a Nigerian teaching hospital

61 Volume 16 No. 2 April 2010 SAJP Depression is twice as likely to occur in individuals with diabetes mellitus (DM) compared with apparently healthy controls. Prevalence rates vary from 11% to 60%. Higher rates are observed depending on study setting (clinical v. community), assessment tool (self-report questionnaire v. diagnostic tool) and design (uncontrolled v. controlled). Depression negatively affects quality of life, treatment outcome and medication adherence of patients with DM. Female gender, low socioeconomic class and the presence of other physical illnesses are associated with an increased likelihood of depression in patients with diabetes.

group, clinical details such as type of diabetes, duration of illness, modality of treatment and presence of other physical co-morbidities were either extracted or corroborated from patient records.
• The depression module of the Schedule for the Clinical Assessment in Neuropsychiatry (SCAN) 13 to diagnose depression.Recorded data were entered into a computer algorithm that generated a diagnosis according to the International Classification of Diseases (ICD-10) criteria.The SCAN has good psychometric properties and has been used in populations in general hospital and psychiatric settings in Nigeria. 14The Beck Depression Inventory (BDI), 15 a 21-item self-report questionnaire for the measurement of depression and its severity.Each item has a score-range of 0 -3; item scores are added to determine the intensity of depression.A minimum total score of 10 is required to identify subjects with significant symptoms.A score range of 10 -18 indicates mild/moderate depression; 19 -29 moderate/severe; and 30 -63 extremely severe depression.The BDI has been standardised and widely used in Nigeria. 15e SCAN and the BDI were translated into pidgin English, which is the lingua franca in the cosmopolitan city of Benin, using the method of back translation.One of the authors administered the questionnaires to respondents to aid consistency.

Ethics
The ethics review committee of UBTH reviewed and approved the study protocol.

Procedure
Patients attending scheduled clinic visits, who satisfied the inclusion criteria, were approached by one of the authors.The nature and purpose of the study was explained and full confidentiality assured.Written informed consent was subsequently obtained from patients who agreed to participate.The socio-demographic questionnaire and BDI were self-administered among literate patients.In illiterate patients, statements on the socio-demographic questionnaire and the BDI were read out, either in English or in the translated pidgin English format.Subsequently, the depression module of the SCAN was administered by one of the authors to diagnose the presence of depression.A similar procedure was employed among the apparently healthy controls.

Data analysis
Data were analysed using the Statistics for Social Science (SPSS) software, version 11.Descriptive statistics were used to summarise the data.Comparisons of categorical and continuous variables were done using the chi-square and t-tests respectively.
Relationships between continuous variables were conducted using the Pearson correlation analyses.

Results
Over the study period, 211 patients satisfied the study's inclusion criteria and were approached, of whom 200 consented to participate in the study (94% participation rate).The same number (200) of apparently healthy controls was recruited from personnel of 3 local governments.The mean age and SD of the diabetes patients (index group) was 47.1±9.6 years.This was comparable with that of the local government staff (control group) of 46.4±9.2 years (t=0.88,df=398, p=0.379).In the index group, 54% were female, 83.5% were married, 66.5% were employed, and 97.5% were Christian.Respondents in the index group had on average significantly more children (5.4±2.2) compared with the average number of 2.9±2.2among the control group (t=11.28,df=398, p<0.001).Furthermore, respondents in the index group had significantly lower monthly incomes -N28 790±38 264 compared with N68 900±30 386 of the control group (t=-11.61,df=398, p<0.001).
In the index group, a majority were being managed for type 2 DM, and most (69%) reported having other physical co-morbidities.The most common (68.5%) were cardiovascular (hypertension, ischaemic heart disease and cerebrovascular incidents).Most (64.5%) of the patients had been diagnosed with diabetes for less than 5 years.The average duration of diabetes was 5.2±5.2years.Approximately 20% of the patients had a family history of diabetes.Sixty (30%) respondents in the index group met a SCAN diagnosis for a depressive disorder according to the ICD-10, which was significantly higher than the proportion of 9.5% who met a SCAN diagnosis of depression in the control group (χ 2 =26.51, df=1, p<0.001).Of those in the index group who had a depressive disorder, 31 (51.7%)met the SCAN criteria for mild depression, 22 (36.7%) were moderately depressed, and the remainder had severe depression.A comparison of some socio-demographic and clinical characteristics of the index group with the presence or absence of a SCAN diagnosis of depression is shown in Table I.

articles
On the BDI, approximately 77 (39.5%) respondents in the index group had BDI scores >10, compared with 27 (13.5%) in the control group.The mean BDI score of 10.77±8.97 in the index group was significantly higher than the score of 4.74±4.82 in the control group (t=8.37,df=398, p<0.001).Monthly income/ allowance as well as number of children was significantly and negatively correlated with BDI scores.There was a positive correlation between age, duration of diabetes and BDI scores; however, this association was not significantly correlated (Table II).

Discussion
The prevalence of depression (30%) among patients with DM in this study using a diagnostic instrument, was much higher than that reported in an earlier study (9.5%) conducted in this environment and employing similar methodology. 3Although the mean age of the population in the study by Issa et al. 3 was similar to that in our study, our sample had a higher representation of females.Anderson and his colleagues 2 observed in their metaanalysis that depression was more likely to occur in women with diabetes than in men.This reason alone may not be sufficient to articles 63  this.[12] We found a significant correlation between earning power, having more children and depression symptom severity as measured by the BDI.DM is a chronic physical disorder that requires lifestyle alterations and medication adherence for its successful treatment.
Individuals with low earning power face the twin burdens of paying for health care, which in Nigeria is largely out-of-pocket, and meeting the needs of their children.Our study design does not permit interpretations as to causality or the temporal relationships between depression, low earning power and larger number of children; however, it may be safe to say that the presence of these characteristics in persons with diabetes could be a pointer to the presence of depressive symptoms, especially in individuals with poor glycaemic control and medication adherence.
Although diabetes patients who were female, married, or had less than 12 years of education were more likely to have a SCAN diagnosis of depression, we found no statistically significant associations.Our finding contrasts the report from Texas, USA, where female gender and a lower duration of formal education were predictors of depression. 7r study is not without limitations.Firstly, it was conducted at one centre, which limits the generalisation of our findings.Secondly, we did not control for other mental illnesses such as anxiety disorder which are co-morbid with depression in DM.Thirdly, we did not determine the validity of the translated questionnaires as well as the reliability of the self-or interviewer-administered forms.

Conclusion
DM is expected to assume pandemic proportions in sub-Saharan Africa. 10Our study further confirms the impression that depression is highly co-morbid with the disease.The negative effect of depression on achieving good glycaemic control means that physicians need to screen for and manage this disorder to improve not only the quality of life of diabetes patients but also reduce overall treatment costs, which are generally unaffordable by most individuals with the disease.

Table II . Correlation between some socio-demographic and clinical characteristics and BDI score
*Statistically significant.