Morbidity and mortality among diabetic patients admitted to Mulago Hospital, Uganda

Abstract

Objectives

To document the causes of admission, clinical presentation and outcome of patients admitted with diabetes mellitus to our medical wards.

Setting

Medical wards of Mulago Hospital, teaching hospital and national referral for the government of Uganda.

Study Design

Cross-sectional descriptive non-interventional study of diabetic medical admissions.

Results

During the study period 129 (4.2%) patients with diabetes mellitus out of 3103 total medical admissions were admitted. The commonest cause of admission was uncontrolled diabetes (48.3%) but infections were present in 27.7% of all the study patients. The commonest infections were pneumonia (15%) and urinary tract infections (11.8%). Diabetic ketoacidosis (DKA) was a cause of admission in 9.2% of all the study subjects. Glycaemic control was satisfactory among 50.6% (HbA1c less than 7) despite 84.5% of the study subjects being hyperglycaemic at admission (mean random blood sugar 20±9.0 mmol/L). Fifty-point seven of the subjects had long term complications of diabetes at admission with hypertension (53.8%) and peripheral neuropathy (38.3%) being the commonest. There were 13 deaths (10.8%) and 61.5% of the deaths were among patients admitted with infections. The average length of hospitalisation was 9.5±4 days.

Conclusion

The results show that the commonest causes of admission were uncontrolled diabetes and infections. The mortality rate was 10.8%.

Introduction

By the year 2010 the total number of people with diabetes is projected to reach 221 million and the regions with the greatest potential increase are likely to be Asia and Africa where prevalence rates are expected to rise 2–3 times its current estimated values¹.

In Africa, diabetes mellitus is a major cause of hospitalization and an important contributor to premature adult mortality^2,3 However, over the subsequent years, the burden of the disease has steadily increased.

In Uganda, no prevalence studies have been done but hospital based evidence suggests increasing numbers of people with diabetes⁴.

Previous studies in Mulago hospital have showed increasing numbers of registered cases from 1414 in 1982, 1853 cases in 1994 and 4937 by April 1998. This is an increase of more than 100% in 4 years. Between June 1999 and June 2000, the number of clinic attendees had increased to 4474, an average of 102 patients per clinic visit. Of these 368 were new patients, with an average of eight new patients per clinic day⁴.

Many retrospective studies in Africa have highlighted the proportion of diabetic admissions to total medical admissions namely, 7% in Sudan⁵, 1% in Tanzania⁶, 1.5% in S.Africa⁷ 0.4% in Nigeria⁸. Unpublished reports from Mulago Hospital show that diabetic admissions accounted for 3.5% of all medical admissions between Jan–Dec 2000⁹. The main objective of this study was to document clinical presentation and outcome of patients admitted with diabetes mellitus in order to devise means of reducing morbidity, admission rates and mortality of these patients.

Patients and methods

Design

A descriptive cross-sectional non-interventional study was undertaken with follow up from admission to discharge.

Setting

The study was conducted in Mulago Hospital, Uganda's national referral and teaching hospital for Makerere Medical School with a bed capacity of about 1000 beds although actual admissions can go up to 1500 patients (Admission Summaries). The hospital receives diabetic patients from all over the country usually to benefit from the successfully run diabetic clinic, which offers diabetic education, investigation and treatment. Each of the triree general wards has two firms, which admit in turns every 6^th day. The firms admit diabetic patients from the casualty (mainly those with acute life threatening complications) and referrals from private clinics and up country district hospitals though the majority of admissions are from the hospitals own diabetic clinic. A cross-sectional descriptive study was carried out between l^st August and 15^th November 2001. The study was approved by research committees of Makerere Medical School and Mulago Hospital.

Patients

All consenting diabetic patients admitted to the medical wards during the study period were eligible for enrolment. These were identified daily from the admission books by the principal investigator and three diabetic nurse specialists and through direct inquiry among each day's admitted patients. The study objectives and procedures were explained to the patients and those who agreed signed the consent form and were each assigned a study number and recruited.

Procedures

Medical history

A detailed history, relating to the presenting illness and infections, duration of diabetes since admission, and identification of possible complications of diabetes mellitus such as peripheral neuropathy, nephropathy, retinopathy, vascular and autonomic neuropathy was obtained from the patients and entered into a questionnaire.

Examination

A physical examination of all the body systems was done looking for signs of infection, other lesions and complications of diabetes. Measurement was done of weight, height, waist circumference, abdominal (girth) circumference and from these, the BMI (Body Mass Index) and WHR (Waist Hip Ratio) were calculated. The ideal WHR was defined as <0.8 for females and <0.95 for males. Patients whose BMI was <18.5 kg/m2 were considered Underweight, 18.5–24.99 Normal weight, 25–29.99 were over weight. Those with BMI > 30 were considered obese 10. Blood pressure was measured in both supine and erect positions using a mercury sphygmomanometer and the difference if any was noted. High blood pressure was defined as >130/ 85 mmHg in patients aged 65 years and below and >140/90 mmHg for those over 65 years according to the WHO-ISH guidelines¹¹. The skin was examined for boils, ulcers and other lesions. The feet were examined for dryness, fissures, ulceration and presence or absence of arterial pulses. A complete neurological examination was done looking for visual field defects, cranial nerve palsies, and sensory loss (using a tuning fork, small piece of cotton and a small blunt pin). Muscle power was graded according to Standard procedure as grade 0 to 5¹². Reflexes were tested using a patella hammer. Obvious ocular abnormalities such as cataracts were noted.

Laboratory methods

A sample of 5 to 10 mls of blood was drawn from the antecubital fossa of each patient under sterile conditions (after cleaning with povidone iodine) for estimation of, HbAlc (measured using the calorimetric method -Human diagnostics), and serum creatinine (Sigma creatinine No.555-A Kit). Poor glycaemic control was defined as HbAlc >7%. A drop of blood obtained from the fingertip was used to estimate random blood glucose using a glucometer (Life scan). Urine sugar and protein were semi-quantitatively estimated by dipstix method (Nephur-Test®).

Patients were followed up from admission to discharge and clinical end points were discharge and death. The duration of admission was recorded in days.

Statistical methods

The total number of medical admissions was computed from the ward admission books at the end of each admission day, added for the study period, and used to calculate the proportion of diabetic admissions to total medical admissions. Data was analysed by Epi-info (Version 6 Atlanta GA 1994) and SAS statistical programmes for means and associations. The level of statistical significance was taken as p<0.05.

Results

Admissions

A total of 3103 patients were admitted to the medical wards during the study period and out of these, 129 (4.2%) were diabetic. One hundred and twenty (93%) of the diabetic patients consented and were recruited into the study, while 4 (3.1 %) did not consent to participate, 5 (3.9%) could not answer the questionnaire because they were too sick. Of the 120 patients recruited in the study, 57.5% were female and 42.5% were male. Their age ranged from 7 to 80 years with a mean (±SD) age of 49 (±18) years. There was no statistically significant difference between the ages of males and females (p-value 0.6). Forty-nine patients (40.8%) were from rural areas while 47 (39.2%) were from periurban areas and 20% were from urban areas. Most of the patients (80.8%) lived within a distance of 10 km to the nearest health facility. Half of the patients had attained education up to primary school level, 20% secondary school and only 13.4% had attained tertiary education. The prevalence of alcoholic consumption and cigarette smoking was 18.3% and 11.7% respectively and of those who took alcohol, 31.8% in smoked. Fortythree (35.8%) of the participants had been previously admitted due to a Diabetes-related condition and about half of these had repeated admissions. Eighty-four patients (70%) had diabetes for less than 5 years, but forty-nine of these had been diagnosed within the previous 2 months and were considered newly diagnosed. Fourteen patients had been diagnosed over 20 years. The medical history of the study subjects is summarised in table 1.

Table 1.

Medical history of the study participants

Characteristic	Number	Percentage (%)
Place of diagnosis
Mulago	48	40.0
Private clinics	36	30.0
Other hospitals	26	21.7
Health centres	10	8.3
Diabetes treatment at time of admission
Insulin	57	47.5
Oral agents	33	27.5
Insulin and oral agents	5	4.2
Diet alone	2	1.7
Not sure	23	19.2
Ever received diabetes education
Yes	89	74.2
No	31	25.8
Prior diabetes related admissions
None	77	64.2
One	24	20.0
Two	7	5.8
Three and more	12	10.0
Diabetic clinic attendance
Ever attended*	44	36.7
Never attended	76	63.3
Previously diagnosed hypertension
Yes	44	36.7
No	76	63.3
Causes of previous admission (n=42)
Diabetes out of control
(uncontrolled hyperglycaemia)	15	35.7
Newly diagnosed	5	11.9
Malaria	5	11.9
Hypoglycaemia	2	4.8
Diabetic ketoacidosis	2	4.8
Hypertension and other Chronic
Complications	6	14.3
Others	7	16.7

^*54.5% of these had attended the diabetic clinic in the previous 8 weeks

Causes of admission

Table 2 summarises the reasons for admission of the study subjects. The commonest cause of admission was uncontrolled hyperglycaemia (48.3%) triggered usually by infections. Infections accounted for 27.5% of the admissions while acute complications were a reason for admission in 24.2% of which hypoglycaemia was the commonest (14.2%).

Table 2.

Reasons for admission

Condition	Number(%)	Mean duration of admission (days)
Infection	33(27.5)
Pneumonia	18(15)	9.7
Urinary tract infections (UTI)	14(11.7)	6.5
Diabetic foot	5(4.2)	20.6
Malaria	4(3.3)	4.3
HIV/AIDS	4(3.3)	-
Newly diagnosed (Diagnosed within 2 months of current admission)	49(40.8)	6.7
Uncontrolled hyperglycemia	58(48.3)	9.3
Acute complication of diabetes	29(24.2)
Hypoglycaemia	17(14.2)	5.6
Diabetic ketoacidosis	8(6.7)	12.7
(DKA)
Hyper osmolar non ketotic	4(3.3)	3.3
coma (HONK)

Long-term diabetic complications

Figure 1 below shows the complications observed among the study patients. At admission, 56.7% patients had long-term complications of diabetes though these were not necessarily a cause of admission. Hypertension was the commonest complication diagnosed on the wards occurring in 71(59.2%) of the study subjects. However, at the time of recruitment only 64 (53.8%) patients had a significantly raised blood pressure (above 130/85 mmHg in those 65 years and younger and 140/90 mmHg in those older than 65 years respectively) since some patients were already on anti-hypertensive medications. Burning sensation in the feet and loss of ankle jerks were the commonest symptom and sign respectively in those who had neuropathy (38.5%).

Figure 1.

Common long-term complications

Laboratory profile

Glycosuria was found in 95 (79.2%) of the patients and correlated well with random blood sugar levels (Correlation coeff. 0.85). The mean random blood sugar (RBS) was 20.0 mmol/L (±9.0) but ranged from 0.2 to 43 mmol/L. Overall 84.5% of the study participants were hyperglycaemic at the time of admission. HbAlc was measured in 77 (64.2%) of the study participants and of these, 50.6% had values within the acceptable limits (<7%) while 24.7% had levels above 10% hence were considered to be very poorly controlled. Proteinuria was observed in 31 (25.8%) and 16% of these had massive proteinuria (urine protein > 500mg/dl). Patients with massive proteinuria had significantly higher mean blood pressures (165±26.5 mmHg systolic and 100±10 mmHg diastolic) compared to those with minimal proreinuria (123±29 mmHg systolic and 73±17.5 mmHg diastolic) p-values 0.04 and 0.003 respectively.

Out come of admissions

Thirteen deaths (10.8%) occurred among the study participants and 8 (61.5%) of these occurred in patients admitted with infections namely; pneumonia (5), malaria (1), urinary tract infection (UTI) (1), and an HIV related infection (1). Only one (12.5%) of the 8 patients with diabetic ketoacidosis (DKA) died during the admission while 2 other deaths occurred due to pneumonia among newly diagnosed patients.

The mean duration of admission was 9.5±SD days but ranged from 1 to 56 days with patients with diabetic foot infection staying longest (mean duration 20.6 days). However RBS, HbAlc, diabetes duration and presence of infection did not significantly affect the outcome of admission.

Discussion

Diabetes mellitus is a chronic disease associated with both acute and chronic complications that often lead to admission to hospitals. The causes of admission among patients with diabetes have to-date been poorly documented. In this study we determined the clinical presentation and outcome of diabetic patients admitted to the medical wards of Mulago Hospital. We also documented the glycaemic control at admission in addition to assessing the prevalence of both acute and chronic complications of diabetes. In this study, diabetic patients constituted 4.2% all medical admissions during the study period. This proportion is a higher than was observed between January and December 2000, where 3.5% of all medical admissions were due to diabetes^4,9. This study demonstrates that diabetes admission had increased with the admission rate of 4.2% being higher than rates reported elsewhere in Africa^6–8. However even higher diabetic admission rates have also been reported in Libya where diabetic admissions constituted 806 (22%) of 3679 total medical admissions in a University teaching hospital¹³.

The mean age of the patients in this study (49 ±18 years) was higher than in previous studies by Otim and others¹⁴. The patients in the later study were out patients and this might explain the age difference noted and it may also indicate that older patients get more problems that require admission compared to younger ones. Majority of the patients seen in this study were either from a rural (40.8%) or peri-urban (39.2%) area of residence. This probably reflects the socio-demographic characteristics of the population in Uganda but may also reflect the result of rural-urban migration and its attendant changes in lifestyle leading to more diabetic patients in urban areas. This may have consequences as far as access to medical care and obtaining drugs for diabetic treatment, which are easier in urban and peri-urban areas. However, 80.8% of all the subjects were within 10km of a health care giving unit and this would influence referral and access to the national referral and diabetic centre.

Inadequate formal and diabetic education, lower socio-economic status in addition to living in rural or peri-urban areas may have led to the poor control since poverty and lack of formal education are factors that hinder proper control of diabetes¹⁵.

According to this study, patients who had had diabetic education had lower random blood glucose at admission though this was not reflected in better long-term glycaemic control nor in did it lessen the risk of development of complications.

Like in most developing countries, uncontrolled hyperglycaemia was the commonest cause of admission (48.3%), and frequently the trigger of the metabolic decompensation is an infection 16. It could also be argued that patients who had uncontrolled hyperglycaemia were more likely to develop acute infections.

This study also shows that long-term complications were significant causes of admission. The reasons for this may be patients were older and had had diabetes for longer as shown in this study but also it may also be explained by improvement in the Standard of care of diabetes.

Diabetes ketoacidosis occurred in 8 (6.7%) of the study participants. Common precipitants for the development of DKA are usually omission of insulin and intercurrent infections¹⁷. In this study, of the 5 patients with DKA 4 had Urinary tract infections and one had pneumonia.

While the American Diabetic Association recommends admission of only newly diagnosed children and adolescents¹⁸, this was not reflected in this study where many adults with uncontrolled diabetes were admitted without any other indications.

Established complications of diabetes were present in 56.7% of the subjects, and among these, hypertension and neuropathy were the commonest. The prevalence of neuropathy (38.3%) elicited from physical signs was lower than that suggested by symptoms of burning sensation (50.8%) and numbness (63.3%) in the feet. This may mean that the physical signs of neuropathy occur later than the symptoms or that the examination techniques used may not be sensitive enough. Peripheral vascular disease was rather uncommon and occurred in only 6.7% of subjects who had symptoms of claudication and absent dorsalis pedis. This is not surprising because, large vessel disease is thought to be less frequent in Africans and is in keeping with the previouly reported prevalence of 2.6% in Uganda¹⁹. This may be an underestimation as patients with chest pain are not always evaluated for Ischaemic Heart Disease (IHD) moreover myocardial infarction may be silent in diabetic subjects (Geof 1994). Ischaemic heart disease was not particularly looked for in this study. Kabole et al reported micro albuminuria in 69% of the patients attending the MHDC²⁰ whereas in our study we found proteinuria in 31 (25.8%) and 47 (39.2%) subjects had a raised serum creatinine. There was no significant relationship between urine protein and level of serum creatinine as would be expected. This further confirms that serum creatinine is not an accurate measure of kidney involvement although patients with massive proteinuria (>1000mg/dl) were assumed to have diabetic nephropathy. Creatinine assays may also give falsely high results due to interference by high ketone levels²¹. Possible confounding factors included hypertension since the patients with massive proteinuria had a higher mean systolic blood pressure. The overall hypertension rate was 53.8% which is much higher than the rates reported by Otim and others in 1996 where the overall hypertension rate among patients attending the MHDC was 36%^14,22. Hypertension rates in other African countries vary between 37% and 45%, and these are lower than our findings^23–26. The difference may be explained by the fact that those quoted studies used different cut off limits for normal blood pressure (140/90 mmHg).

The mortality rate among our patients was 10.8% which is lower than the 18% reported from Libya. This study and reports from other African countries show how poor diabetic outcomes can be in Africa. In Zimbabwe, 9 (8.4%) of 107 newly diagnosed diabetic patients died during the admission mostly from DKA or hyperosmolar coma²⁷, while in Tanzania 205 (16%) of a cohort of 1250 diabetic patients died26. Infections were the commonest cause of death (61.5%), with pneumonia (38.5%) being the one associated with most deaths and at least one person dying of an AIDS related condition. This contrasts with results of autopsies done in diabetic patients in this same hospital between 1957 and 1966, where two thirds of the deaths among 54 reported autopsies were due to ketoacidosis or hypoglycaemia and the rest (one third) were almost entirely due to infections²⁸. In our study, only one of the patients with DKA died similar to what has been reported in some African series where ketoacidosis was the reported cause of death in 19 (23.8%) of 80 Nigerian patients⁸ and 10% of 100 deaths in Ethiopia¹⁷.

This study shows that uncontrolled hyperglycaemia, infections and chronic complications of diabetes are a major cause of morbidity frequently leading to hospitalisation and that infections are frequent causes of hyperglycaemia, poor control and mortality. Encouraging patients to attend outpatient clinics, national diabetic programmes and the increased use of diabetic nurse educators should improve both the quality and quantity of diabetic education as well as overall care of the patients.