Prevalence, types and factors associated with echocardiographic abnormalities among newly diagnosed diabetic patients at Mulago Hospital.

Background: The prevalence of Diabetes mellitus (DM) is on a rise in sub-Saharan Africa and will more than double by 2025. Cardiovascular disease (CVD) accounts for up to 2/3 of all deaths in the diabetic population. Of all the CVD deaths in DM, 3/4 occur in sub Saharan Africa (SSA). Non invasive identiﬁcation of cardiac abnormalities, such as Left Ventricular Hypertrophy (LVH), diastolic and systolic dysfunction, is not part of diabetes complications surveillance programs in Ugan da and there is limited data on this problem. This study sought to determine the prevalence, types and factors associated with echocardiographic abnormalities among newly diagnosed diabetic patients at Mulago National referral hospital in Uganda. Methods: In this cross sectional study conducted between June 2014 and December 2014, we recruited 202 newly diagnosed adult diabetic patients. Information on patients' socio-demographics, bio-physical proﬁle, biochemical testing and echocardiographic ﬁndings was obtained for all the participants using a pre-tested questionnaire. An abnormal echocardio gram in this study was deﬁned as the presence of LVH, diastolic and/or systolic dysfunction and wall motion abnormality. Bivariate and multivariate logistic regression analyses were used to investigate the association of several parameters with echocardiographic abnormalities. Results: Of the 202 patients recruited, males were 102(50.5%) and the mean age was 46±15 years. Majority of patients had type 2 DM, 156(77.2%) and type 1 DM, 41(20.3%) with mean HbA1C of 13.9±5.3%. Mean duration of diabetes was 2 months. The prevalence of an abnormal echocardiogram was 67.8 % (95% CI 60%-74%). Diastolic dysfunction, systolic dysfunction, LVH and wall motion abnormalities were present in 55.0%, 21.8%, 19.3% and 4.0% of all the participants respectively. In bivariate logistic regression analysis, the factors associated with an abnormal echocardiogram were age (OR 1.09 [95% CI 1.06–1.12], P <0.0001), type 2 DM (OR 5.8[95% CI 2.77-12.07], P<0.0001), hypertension (OR 2.64[95% CI 1.44-4.85], P=0.002), obesity (OR 3.51[955 CI 1.25-9.84], P=0.017 and increased waist circumference (OR 1.02[95% CI 1.00-1.04], P=0.024. On Multiple logistic regression analysis, age was the only factor associated with an abnormal echocardiogram (OR 1.09[95%CI 1.05-1.15], P<0.0001). Conclusion: Echocardiographic abnormalities were common among newly diagnosed adults with DM. Traditional CVD risk factors were associated with an abnormal echocardiogram in this patient population. Due to a high prevalence of echocardiographic abnormalities among newly diagnosed diabetics, we recommend screening for cardiac disease especially in patients who present with traditional CVD risk factors. This will facilitate early diagnosis, management and hence better patient outcomes. types and factors associated with echocardiographic among


Introduction
Diabetes is an important cause of mortality and morbidity worldwide 1 . The worldwide prevalence of diabe-between 8% and 13% in more developed areas such as South Africa and in populations of Indian origin 5 . An aging population together with rapid urbanization will lead to an increase in the prevalence of diabetes by the year 2025 6 at which time the majority of the world's diabetespopulation will be living in the developing countries 7 . By the year 2025, the prevalence of diabetes in sub-Saharan Africa is expected to more than double the current figures (3).
Cardiovascular diseases, the commonest cause of death in patients with diabetes, accounts for up to two-thirds of all deaths in this group 5 . The risk of heart disease in diabetes is increased up to 15 years prior to the diagnosis of diabetes 8 . Management of established CVD in these SSA countries is unrewarding due to chronic shortage of resources. Early diagnosis of preclinical CVD in this patient population is therefore a priority as it will enable early intervention to prevent or delay complications. Subclinical cardiac abnormalities of structure and function, detectable using myocardial Tissue Doppler Imaging (TDI) and conventional echocardiographic imaging are common in adults with diabetes 9 . With increasing availability of this non-invasive cardiac ultrasound service including point-of-care hand-held echocardiographic machines, it is reasonable to argue that echo screening in all diabetes be part of routine care.
The absence of systematically collected data on pre-clinical CVD in these patients makes it impossible to implement policy change as suggested above as no studies have been carried out to assess the magnitude of the problem in this patient group.
Accordingly, the aim of this study was to determine the prevalence, types and factors associated with echocardiographic abnormalities among newly diagnosed diabetic patients at Mulago hospital.

Study design
This was a cross-sectional study among 202 newly diagnosed diabetic patients at Mulago National referral hospital in Uganda conducted from June to December 2014. The study was carried out in the diabetic outpatient clinic and the medical wards; 4B endocrine and 3BEM of Mulago National referral Hospital. Mulago serves as the National referral hospital for Uganda and Teaching Hospital for Makerere University with a bed capacity of 1500. A diagnosis of DM was made by one of the following: Fasting glucose >126mg/dL, 2 hour plasma glucose of >200mg/dl after oral Glucose tolerance test (OGTT), or glycated hemoglobin (HbA1C) >6.5%. We differentiated type 1 from type 2 DM using clinical assessment because we were unable to assess insulin levels which is the gold standard. Blood pressure was measured using a mercury sphygmomanometer and an average of the 2 readings was used in the analysis. Hypertension was defined as present if subjects were on anti-hypertensive medication, had history of hypertension and/or evidence of hypertension (blood pressure ≥140/90 mmHg).
All newly diagnosed diabetic patients aged 18 years and above attending the diabetic clinic or admitted to the medical wards of Mulago hospital during the study period who met the inclusion criteria and provided informed consent were recruited consecutively. Using prevalence of 72% for an abnormal echocardiogram among diabetics as determined by Piyush 10 , a sample size of 309 was estimated using the Kish Leslie (1965) formula 11 . However we were unable to achieve the required sample size due to limitations in logistics for the investigations.

Operational definitions:
Type1 DM: This referred to patients who required insulin and were depending on insulin for glucose control since diagnosis. Type 2 DM: Patients were classified as having type 2 diabetes mellitus if they required oral hypoglycemic agents or usage of combination of insulin and the oral hypoglycemic agents for glucose control. Type 3 DM is diabetes secondary to an identifiable cause like pancreatitis or cancer of the pancreas. Type 4 DM is gestational diabetes. Fig. 1 illustrates the patient recruitment flow.
The following study parameters were collected using a pre tested questionnaire: patients' demographic data, history of hypertension, age, physical exercise, marital status, date of diagnosis of DM, drug history, occupation, education level, LNMP, Blood pressure, body Weight, height, Waist and hip circumferences. Glycated haemoglobin (HbA1C) was measured by automated high performance liquid chromatography. Other investigations included urinalysis and Microalbuminuria using ACR. Echocardiography parameters were acquired using a commercially available machine, Phillips HD11XE (Eindhoven, The Netherlands) with 2-D, M Mode and Doppler capabilities was used according to the American Society of Echocardiography 12 .

Ethics approval
The study was approved by the School of Medicine Research and Ethics Committee of the College of Health Sciences, Makerere University. Written informed con-sent for involvement in the study was obtained from the study participants. Enrolment was totally free and voluntary and participants were free to withdraw at any time without any consequences to them. The patients' records/information was anonymized and de-identified prior to analysis.

Data management and analysis
Data was captured on hard copy Case Report Forms, then transferred to the electronic EPIDATA version 3.1 and exported to STATA 13 for analysis. Continuous variables were summarized as mean (± standard deviation), median (± inter quartile range) and presented in tables. Categorical data was summarized using frequency and percentages and results are presented in frequency tables and bar charts. Logistic regression was used to determine odds for a given outcome and presented as Odds ratio (OR) and their 95% confidence interval (CI). Only factors with P-value <0.2 at bivariate anal-ysis were considered for multivariate analysis. Multivariate logistic regression was used to determine factors associated with an abnormal echocardiogram and the different echo abnormalities (secondary outcomes): significance was at 5%.

Results
This study recruited 202 newly diagnosed diabetic patients between May 2014 and January 2015. Of these, 102(50.5%) were males. The mean age of the participants was 46 years . Majority of patients had type 2 DM, 156 (77.2%) or type 1 DM, 41 (20.3%). None of the patients had type 3 DM, and the remainder had ges-tational DM, 5 (2.5%). The mean HbA1C was 13.9%, (SD: 5.27%). Mean duration of diabetes was 2 months. Majority of patients 128 (63.6%) were unemployed. The education levels of participants were mainly primary which accounted for 78 (38.6%). The remainder of patients had had either secondary education 75 (37.1%), tertiary education 31 (15.4%), or no formal education at all 18 (8.9%). Majority of the study participants 118 (58.4%) were married. Study participants who had never been married were 30 (14.9%) while those who were once married and no longer married were 54(26.7%). Of all the participants, 125(61.95%) had hypertension according to the operational definition and only 38.1% were normotensive.

Echocardiographic abnormalities.
Transthoracic echocardiography was performed on all 202 participants and the results are shown in Table 2.
Prevalence of an abnormal echocardiogram was 67.8 % (95% CI: 60%-74%). Isolated Diastolic dysfunction was present in 55.0% of all the participants (95%CI:  Generally, the prevalence of the different echocardiographic abnormalities was much higher among patients with type 2 compared to type 1 DM (Figure 2). Of these echocardiographic abnormalities 120 (87.6%) were seen among type 2 Diabetics while type 1 DM accounted for 15 (11.0%). Wall motion abnormalities were only present among type 2 DM patients; there were no patients with type 1 DM with wall motion abnormalities ( Figure  2).
After adjusting for patients' sex, hypertension, BMI, waist hip ratio and waist circumference the only significant factor associated with an abnormal echocardiogram was age (OR1.09 [1.05-1.15] P<0.0001). Hypertension, obesity, waist hip ratio and waist circumference were no longer significantly associated with an abnormal echocardiogram in this model. their study assessed 141 type 1 diabetic patients and found the prevalence to be 2.5 times higher than what we found. The major abnormalities on the echocardiogram was diastolic dysfunction, LVH and systolic dysfunction 14 . The difference in the studies is that the latter looked at patients with an average duration of diabetes of 21 years and only type 1 DM, yet in our study we enrolled only newly diagnosed diabetics with mean duration of DM of only 2 months.

Diastolic dysfunction
In Africa, Diabetes-related cardiovascular disease complications were considered to be rare but are on the rise 5 . We found the prevalence of LVDD to be 55% in our study. This is in keeping with earlier studies in Africa which showed that Diabetic cardiomyopathy, which is largely represented by diastolic dysfunction, may affect up to 50% of all patients 5 .
Over the last three decades several studies have demonstrated the presence of left ventricular (LV) diastolic dysfunction in patients with normal LV ejection fraction (LVEF) 10,14-17 . Majority of these studies have found a high prevalence of Left ventricular diastolic dysfunction ranging from 40% to 75% 10, [14][15][16]18 .
The prevalence of LVDD in patients with T2DM has been shown to vary from 25% to 60% in different studies [19][20][21] . It is noted that majority of patients in our study had type 2 DM and were newly diagnosed as cited earlier. This could account for the similarity in the results.
In Senegal LVDD has been found in 62% of type 2 DM patients 22 .
Prior to the development of symptomatic heart failure, subclinical LV dysfunction (systolic and diastolic) may exist for some time and are associated with a poor prognosis 13,15 . We used mitral E/A ratio to define diastolic dysfunction unlike in the studies quoted where they used both E/A and TDI for diastolic dysfunction a method that increases the sensitivity in diagnosis. We used mitral E/A because it is the standard of care in our setting and TDI has not been widely used. This means that the prevalence of diastolic dysfunction we reported of 55% among newly diagnosed diabetics may be an under estimate of a bigger problem. Omission of TDI may eliminate up to 25% of patients with pseudonormalization which may go unmasked 10 . This may call for longitudinal studies to evaluate this problem further using TDI.
The clinical implication of this diastolic dysfunction which was found in majority of our patients is the inevitable progression to clinical heart failure. In fact Progression of diastolic dysfunction to heart failure is 4 years in type 2 DM 9,10 and 6 years in type 1 DM 14 . There is great need to follow up these patients before overt clinical outcomes occur.

Left ventricular hypertrophy (LVH).
From a recent cohort in Australia among type 1 DM patients 10 , the prevalence of LVH alone was 9%, which is twice the prevalence in the general population; however, in our study, we found a higher prevalence up to 19.3%. This could be explained by the high prevalence of hypertension (62%), which was strongly associated with LVH in our study.
Valensi, in an African population, found LVH in 33.6% of the study population 17 . The population he assessed had one or more CVD risk factors and used both rest echocardiography and myocardial scintigraphy which could have led to a higher prevalence compared to what we found. Piyush in a cohort among type 2 DM patients found LVH in 51% of all patients 10 . The difference between his and our findings may be due to a long duration of DM of 10 years in his study while our patient population was newly diagnosed. Furthermore, our patients were younger; mean age 46 years compared to Piyush's cohort which had 61 years as their mean age. Screening for left ventricular hypertrophy in patients with type 2 diabetes mellitus in the community revealed a prevalence of up to 56%. This prevalence was much higher than what we found, possibly because our patient population was newly diagnosed and the duration of DM was only 2 months 23 . In the former, patients had a minimum duration of diabetes of 5 years. This may call for longitudinal studies to assess the burden of LVH and its association with duration of diabetes in our setting.

Systolic dysfunction
We found a high prevalence of systolic dysfunction at 21.8% among newly diagnosed diabetic patients in this study. Majority of these patients 37(84.1%) had type 2 DM. Piyush in Australia found a lower prevalence of systolic dysfunction among type 2 DM patients which was present in 5% 10 . This is greatly different from what we found yet patients in Australia were much older; mean age 65 compared to 46 and duration of DM of 10 years compared to 2 months in our study. This calls for more studies to clear the difference.
There is evidence of high prevalence of reduced LVEF in diabetic patients without known heart disease. In one study the prevalence of reduced LVEF was 16.7% 24 , comparable to our findings. Valensi in an African population found a lower prevalence of systolic dysfunction of 3.2% compared to what we found. The reason for the difference could be the fact that all patients assessed had no clinical features of heart failure in Valensi's cohort 17 . In our study we did no clinical assessment. We found that 27 (13.4%) of all patients had combined systolic and diastolic dysfunction in our study. This finding is similar to what was found in a cohort in Australia which found combined systolic and diastolic dysfunction at 11% 10 . Both studies assessed type 2 diabetic patients.
Wall motion abnormalities Regional wall abnormalities were present among 8 (4.0 %) of all the study participants in our study. In Africa, the findings have been similar to what earlier studies in the developed world have shown 15,17,22 . Valensi generated evidence from an African diabetic population: Rest echocardiography revealed that hypokinesia was found in 6.1% of the study population 17 . This finding is comparable to what we revealed in this study. The slight difference may be due to the fact that in the former study, both rest echocardiography and myocardial scintigraphy were used; hence a higher sensitivity.
Factors associated with echocardiographic abnormalities among newly diagnosed diabetic patients at Mulago Hospital.
In this study, we found that traditional CVD risk factors such as advancing age, hypertension, obesity, increasing waist circumference and type 2 DM were associated with an abnormal echocardiogram in newly diagnosed adult diabetic patients in Uganda (in the unadjusted model).
These findings are similar to what was found in other previous studies on echocardiographic abnormalities among diabetic patients 10,14 . The increasing prevalence of factors like hypertension and obesity has led to an increase in the risk and prevalence of cardiac abnormalities in DM 25 . This is in line with what we are seeing in this study.
We found a high burden of hypertension at 62% among the study population. It was strongly associated with an abnormal echocardiogram in our study. This is in agreement with earlier studies which described that hyper-tension is more prevalent in diabetic patients in sub-Saharan Africa 26 . The other reason which can explain this association is that: Individuals with hypertension are at increased risk of DM compared with normotensive persons. Furthermore, up to 75% of cases of cardiac disease in DM can be attributed to hypertension 27 .
Piyush, in a longitudinal study in Australia among type 2 DM patients found the independent predictors of an abnormal heart echocardiogram were obesity, age and the number of antihypertensive agents used. Age was the strongest predictor of an abnormal echocardiogram, with the risk increasing by 9% for each year over 50 years of age 10 . This is similar to what we found that every 1 year increase in age has a 9% increased odds of having an abnormal echocardiogram.
After multivariate adjustment, however, only age was independently associated with echocardiographic abnormalities among newly diagnosed diabetic patients.
Life expectancy in sub-Saharan Africa; including Uganda has risen in the past 50 years. Many more people living with DM are therefore exposed to the CVD risk factors for long periods for the cardiac complications to develop 5 .This may be the reason why age is an important factor in this study. In addition, advancing age is a known risk factor for diastolic dysfunction in the general population; this could have contributed to the high prevalence among the elderly.
Gough described that glycemic control does not affect cardiac abnormalities in the short term among diabetics 26 . We reached a similar conclusion in this study: glycemic control was not associated with cardiac abnormalities despite very profound hyperglycemic states, mean HbA1C was 13.4%. Similarly microalbuminuria measured by urine ACR was not associated with an abnormal echocardiogram, despite the fact that significant microalbuminuria was present in 54% of all patients. This is in contrast with the Strong Heart Study in diabetes, which found urinary albumin excretion was independently associated with cardiac dysfunction 28 . In the later study ACR was measured at least twice but in our study only one measurement was taken.

Limitation of the study
We recognize that the lack of an non diabetic control population is a limitation of our study; however, this study was not designed to determine the effect of diabetes on cardiac structure and function, but rather to explore the magnitude of subclinical cardiac abnormalities among newly diagnosed diabetic patients in the SSA context, including the risk associated with traditional CVD risk factors. Secondly, we used spot urine ACR as a single measure yet persistent microalbuminuria would be diagnosed by at least 2 measurements depicting 20mg/g according to our laboratory. This was due to logistical limitations.
We measured diastolic dysfunction using mitral E/A ratios because it is the standard of care in our setting. Although rigorous TDI in addition to E/A ratios from conventional Doppler studies was necessary to unmask pseudonormalization, it was not done due to lack of capacity. This may result in missing up to 25% of cases of diastolic dysfunction with pseudonormalization. Therefore the prevalence of diastolic dysfunction we are reporting may be an under estimate of a bigger problem. Patients with clinical heart failure were not excluded and this may confound our findings because patients with heart failure may develop diabetes and conversely diabetics are at risk of heart failure. This is one of the limitations in this study. We differentiated type1 and type 2 DM using clinical assessment because we could not assess insulin levels which is the gold standard: this is a limitation for this study. We were unable to enroll the desired number of participants according to the sample size estimated. This was due to limitations in logistics for the required investigations.

Conclusion
Overall prevalence of an abnormal echocardiogram among newly diagnosed diabetic patients was high at 67.8% with majority having diastolic dysfunction as the main abnormality. Left ventricular hypertrophy, systolic dysfunction and regional wall abnormalities were common findings in this study. Traditional CVD risk factors including: Advancing age, hypertension, type 2 DM, obesity and increasing waist circumference were associated with an abnormal echocardiogram in this population. Microalbuminuria, a known CVD risk factor measured using urine ACR was not associated with any of the echocardiographic abnormalities despite a high prevalence at 54%. Similarly HbA1C, which is a measure of glycemic control, was not associated with echocardiographic abnormalities although this patient population had profound hyperglycemic states: mean HbA1C was 13.4%.

Acknowledgement.
Research reported in this manuscript was supported by the Fogarty International Center of the National Institutes of Health under Award Number R24TW008861. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. The authors are grateful to the following persons for their invaluable support: Prof. Nelson Sewankambo, Prof. Moses R. Kamya, the staff of ward 4B endocrine, diabetic clinic, the echocardiography and clinical laboratory of Mulago hospital.