Prevalence of diabetic retinopathy, cataract and visual impairment in patients with diabetes in sub-Saharan Africa
- Simon J Glover1,
- Philip I Burgess2,
- Danielle B Cohen1,
- Simon P Harding2,
- Helma W C Hofland1,
- Eduard E Zijlstra1,
- Theresa J Allain1
- 1Department of Medicine, College of Medicine, University of Malawi, Blantyre, Malawi
- 2Ophthalmology Research Unit, University of Liverpool, UK
- Correspondence to Philip I Burgess, Ophthalmology Research Unit, University of Liverpool, Royal Liverpool University Hospital, Liverpool L7 8XP, UK;
- Accepted 20 March 2011
- Published Online First 21 April 2011
Background/aims There are few published data on the prevalence of diabetic retinopathy in sub-Saharan Africa. We report the prevalence of all grades of retinopathy and associations with systemic parameters in patients attending a secondary care diabetes clinic in Blantyre, Malawi.
Methods Cross-sectional study of all patients attending for diabetes care in a hospital setting. Clinical examination and biochemical testing was performed to assess visual acuity (VA), grade of retinopathy (slit lamp biomicroscopy), microvascular complications, glycaemic control, hypertension and HIV status. Sight-threatening diabetic retinopathy (STDR) was defined as moderate preproliferative retinopathy or worse, circinate maculopathy or exudates within one disc diameter of the foveal centre or clinically significant macular oedema.
Results In patients with type 2 diabetes (n=249) the prevalence (95% CI) of any retinopathy, STDR and proliferative diabetic retinopathy (PDR) was 32.5% (26.7 to 38.3%), 19.7% (14.7 to 24.6%) and 4.8% (2.2 to 7.5%), respectively. The presence of STDR was associated with albuminuria (OR 2.6; p=0.02), the presence of neuropathy (OR 3.4; p=0.005) and insulin use (OR 5.3; p=0.0004), but not with HIV status. In patients with type 1 diabetes (n= 32), the prevalence of any retinopathy, STDR and PDR was 28.1% (12.5 to 43.7%), 18.8% (5.2 to 32.2%) and 12.5% (1.0 to 24.0%), respectively. 12.1% of study subjects had VA worse than 6/18 (20/60).
Conclusion This study provides baseline information on prevalence of all grades of retinopathy and STDR in consecutive cases attending an urban/semi-urban diabetes clinic in sub-Saharan Africa. Prevalence of STDR was high and in type 2 diabetes was associated with albuminuria, neuropathy and insulin use.
- Diabetes complications
- diabetes mellitus
- diabetic retinopathy
- sub-Saharan Africa
The prevalence of diabetes, particularly type 2 diabetes, is rising in epidemic proportions worldwide. Diabetic retinopathy (DR) is the fifth leading cause of global blindness, affecting an estimated 1.8 billion people, and is responsible for 4.8% of blindness.1 Laser photocoagulation has been shown to be effective at reducing the likelihood of visual impairment and blindness in patients with proliferative diabetic retinopathy (PDR)2 and macular oedema3 if timely treatment is performed.
The prevalence and incidence of sight-threatening diabetic retinopathy (STDR) in developed countries have been well documented.4–6 Associations between systemic factors, including poor glycaemic control7 8 and hypertension,9 and the development and progression of retinopathy in these populations are well known. In contrast there are a paucity of published data on DR in sub-Saharan Africa. Only four studies have been published10–13 and none from outside South Africa. In this resource-poor setting, high rates of malnutrition, infectious disease, HIV and anaemia are likely to affect the spectrum of pathology encountered. In order to estimate the current and future burden of disease, and provide data to enable the assessment of changes that may result from service development, accurate data on prevalence and incidence rates are urgently needed.
Queen Elizabeth Central Hospital (QECH) in Blantyre is the only teaching hospital in Malawi, a small country in south-east Africa. It provides primary and secondary care to the population of greater Blantyre (approximately 1 million people, 50% adult), an urban and semi-urban population, and tertiary care to the southern region of the country (approximately 5.9 million people). The recent WHO Malawi national STEPwise approach to surveillance (STEPS) survey estimated a prevalence of diabetes of 5.6% in adults aged 25–64 years.14 The diabetes clinic at QECH represents the most specialised care in the public sector available to patients with diabetes in Malawi. At the time of the study there was no publically provided service for the management of DR including for laser photocoagulation.
We performed a cross-sectional study to investigate the prevalence of grades of DR in adult patients attending the diabetes clinic at QECH. Secondary aims were to characterise associations between systemic characteristics and STDR, and to document the visual acuities (VA) of study subjects.
Materials and methods
A cross-sectional survey was performed in the diabetes outpatient clinic at QECH, Blantyre, Malawi. Consecutive adults attending for routine care between March 2007 and June 2007 were invited to join a prospective cross-sectional study.
Best corrected VA was measured using a Snellen chart. Pupils were dilated with 1% tropicamide±10% phenylephrine. Slit lamp biomicroscopic retinopathy grading with 90 and 60 D lenses was performed by one experienced ophthalmologist (SJG). Adaptation of the Early Treatment Diabetic Retinopathy Study (ETDRS) grading15 for both retinopathy and macular exudates as used in the Liverpool Diabetic Eye Study has been described previously16 (table 1). Macular oedema was assessed according to the ETDRS criteria for clinically significant macular oedema (CSMO).15 STDR was defined as any of the following: moderate preproliferative retinopathy or worse (level 40–71+: sight-threatening retinopathy), circinate maculopathy or exudates within one disc diameter of the foveal centre, or CSMO (level 3–4: sight-threatening maculopathy). Cataract was graded according to the lens opacities classification system III (LOCS III)17 and considered clinically significant when graded at ≥3 in any category (nuclear opalescence, nuclear colour, cortical or posterior subcapsular).
This study of DR formed part of a larger survey of complications of diabetes: full details of the systemic evaluation have been described by Cohen et al.18 Briefly, information on demography and management of diabetes was obtained by a specifically designed study questionnaire and by reference to the ‘health passport’ carried routinely by all patients in Malawi. The classification of diabetes by type was hampered by poor previous history and absence of systematic records. Diabetes with young age at onset and early use of insulin was deemed to be type 1 with all others as type 2. Patients with type 2 diabetes were subclassified based on treatment: insulin-requiring with or without oral hypoglycaemic agents, oral hypoglycaemic agents alone or dietary measures alone.
Physical examination was carried out by a trained clinician who was a member of the study team. Neurovascular foot assessment was performed using an abridged 4-point monofilament examination. Measurements of fasting blood glucose and glycosylated haemoglobin (HbA1c) (Micromat II system; Bio-Rad, Richmond, CA, USA) were recorded. All patients were tested anonymously for HIV using two standard rapid immunoassays (Uni-Gold Recombigen HIV, Trinity Biotech, Bray, Ireland and Determine HIV-1/2, Inverness medical, Princeton, NJ, USA). A subset of patients were tested for albuminuria (Alburtis; Bayer, Tarrytown, New York, USA) and serum creatinine level.
Calculations were performed using Statistical Package for Sociological Sciences (SPSS) software. For the purposes of this study, rates of grades of DR were calculated by the patient using the worse or only gradable eye. VA data were investigated by eye and by patient (better eye). 95% CIs were calculated for proportions. Univariate and multivariate logistic regression was used to investigate the independence of the association of the presence of STDR with duration of diabetes, systolic and diastolic blood pressure, albuminuria, raised creatinine, presence of neuropathy, insulin treatment, sex, current age, fasting glucose, HbA1c, HIV status and body mass index (BMI). Adjusted ORs and 95% CI for presence of STDR were calculated. All tests were two-sided and data were considered significant when p<0.05.
Two hundred and eighty-one subjects were included in the study: 32 with type 1 diabetes and 249 with type 2 diabetes. For those with type 2 diabetes, age at examination (median (IQR)) was 56.4 (49.1–63.7) years, duration of diabetes was 5.2 (2.2–9.0) years and 27.2% were male. Some (19.2%) were taking insulin, 76.0% were taking oral hypoglycaemic agents alone and 4.8% were on diet alone. Age at examination of type 1 patients was 34.1 (26.0–44.3) years, duration of diabetes was 6.7 (3.7–11.4) years and 53.1% were male. Most patients with diabetes had undergone screening for renal disease: 87.1% (n=217) with type 2 diabetes and 87.5% (n=28) with type 1 diabetes.
Prevalence of retinopathy in patients with type 1 diabetes
The overall prevalence (with 95% CI) of retinopathy grades is shown in table 2. The prevalence of any retinopathy, STDR and PDR in patients with type 1 diabetes was 28.1% (12.5 to 43.7%), 18.8% (5.2 to 32.2%) and 12.5% (1.0 to 24.0%), respectively. Four eyes were deemed ungradable due to corneal or lenticular opacities. For one patient both eyes were ungradable. Associations of STDR in type 1 subjects (OR with 95% CI) are shown in table 3. Presence of neuropathy was significantly associated with STDR in univariate analysis.
Prevalence of retinopathy in patients with type 2 diabetes
The prevalence (with 95% CI) of any retinopathy, STDR and PDR in patients with type 2 diabetes was 32.5% (26.7 to 38.3%), 19.7% (14.7 to 24.6%) and 4.8% (2.2 to 7.5%), respectively. Fifteen eyes were deemed ungradable. In four patients both eyes were ungradable. Prevalence of any retinopathy, STDR and PDR increased with time since diagnosis of diabetes (figure 1). For patients with duration of diabetes of ≤5 years, prevalence of any retinopathy was 23.2% (15.8 to 30.6%), STDR 11.2% (5.7 to 16.7%) and PDR or worse 3.2% (0.1 to 6.3%). In patients with disease duration of ≥16 years, prevalence of any retinopathy was 66.7% (46.5 to 86.8%), STDR 42.9% (21.7 to 64.0%) and PDR or worse 14.3% (−0.6 to 29.2%). Patients with type 2 diabetes treated with insulin had a higher prevalence of STDR than non-insulin treated patients. The overall prevalence of STDR was 42.9% (29.0 to 56.7%) in insulin-treated patients compared with 14.0% (9.2 to 18.8%) in individuals not treated with insulin.
Associations of STDR in patients with type 2 diabetes (OR with 95% CI) are shown in table 4. Duration of diabetes, systolic blood pressure, albuminuria, raised creatinine, presence of neuropathy and insulin treatment were associated with STDR in univariate analysis. Albuminuria, presence of neuropathy and insulin treatment were risk factors for STDR in multivariate analysis. Sex, current age, diastolic blood pressure, fasting blood glucose, HbA1c, HIV status and BMI were not significantly associated with STDR.
Presence of cataract and other pathology
In patients with type 1 diabetes clinically significant cataract was present in both eyes in 9% of subjects and in one eye in 3%. In those with type 2 diabetes clinically significant cataract was present in both eyes in 12% of subjects and in one eye in 3%. Three eyes were noted to have central retinal artery occlusions and one eye a central retinal vein occlusion, all in patients with type 2 diabetes.
Frequency distributions of visual acuity (VA) (with 95% CI) are shown in table 5. According to WHO definitions 239 subjects (85.0% (80.9 to 89.2%)) had normal vision (6/6–6/18), 24 subjects (8.5% (5.3 to 11.8%)) were visually impaired (<6/18–6/60) and 10 subjects (3.6% (1.4 to 5.7%)) were severely visually impaired or blind (<6/60 – no perception of light (NPL)). Of subjects in the last category, in the opinion of the examining ophthalmologist, visual impairment was due to DR in six out of 10 cases. Visual impairment and severe visual impairment or blindness increased with age (figure 2). In univariate analysis vision worse than 6/18 was significantly associated with STDR (OR 5.13; 95% CI 2.4 to 10.9; p=0.00002), presence of clinically significant cataract in both eyes (OR 11.7; 95% CI 5.0 to 27.0; p=0.00001) and increasing age (OR 1.077/year; 95% CI 1.038 to 1.118; p=0.0001) but not duration of diabetes (OR 1.047/year; 95% CI 0.994 to 1.102; p=0.081).
We report prevalence rates of DR in patients attending for routine primary and secondary diabetes care in sub-Saharan Africa. This is the first such study from this region outside South Africa. The prevalence of any retinopathy, STDR and PDR in type 2 diabetes was 32.5%, 19.7% and 4.8%, respectively and in type 1 diabetes 28.1%, 18.8% and 12.5%, respectively. Prevalence of sight-threatening maculopathy was 14.5% in patients with type 2 diabetes and 18.8% in those with type 1 diabetes. In multivariate analysis albuminuria, presence of neuropathy and insulin treatment were significantly associated with STDR in patients with type 2 diabetes.
The prevalence of retinopathy in our study is comparable to similar reports from South Africa that were carried out in urban populations from predominantly low socioeconomic backgrounds. However, in contrast to our study, a significant proportion of subjects of European (up to 60%11) and Asian (up to 12%10) descent were included. Only two studies were performed in specialist diabetes clinics.10 11 Carmichael et al10 graded fundus photographs of 1517 patients with diabetes reporting prevalence of any retinopathy 26.5% and 12.6% as ‘severe retinopathy’ (defined as exudates within two disc diameters of the foveal centre or intraretinal microvascular abnormalities (IRMAs), or venous beading, and therefore equating to STDR). Kalk et al11 reported photographic screening of 507 patients, predominantly with type 2 diabetes. The overall prevalence of ‘severe retinopathy’ (again equivalent to STDR) was 13%, although this rose to 19% in black African patients. In 400 subjects Mash et al12 reported prevalence of severe non-proliferative DR of 22.1%, PDR 6.1% and any maculopathy 15.2%. Longer duration of diabetes in this cohort (mean 7.4 years) could explain higher prevalence compared with our study. Read and Cook13 reported 8.9% prevalence of sight-threatening retinopathy in 248 patients with type 2 diabetes. The use of direct ophthalmoscopy in this study, a method known to have a low sensitivity for detection of STDR,16 may account for the low rate.
Prevalence of DR has been more comprehensively studied in Western populations. The Wisconsin Epidemiologic Study of Diabetic Retinopathy (WESDR) reported the prevalence of any retinopathy in patients with type 2 diabetes of 39% in non-insulin users and 70% in insulin users.6 More recent population-based studies in Europe have reported lower rates than WESDR,4 19–23 in part reflecting changes in diabetes management over this time. Younis et al4 studied 8062 patients with diabetes entering an English primary care-based screening programme. The prevalence of any retinopathy, STDR and PDR was 45.7%, 16.4% and 3.7%, respectively, in type 1 diabetes and 25.3%, 6.0% and 0.5%, respectively, in type 2 diabetes. The higher prevalence of retinopathy in our study and those from South Africa compared with recent Western studies is probably multifactorial in origin. It is likely to reflect factors including ethnicity, late diagnosis, poor access to medical services and co-pathology including hypertension, malnutrition, malaria and anaemia.
In our present study, in patients with type 2 diabetes neuropathy and albuminuria were significantly associated with STDR. In the UK Prospective Diabetes Study (UKPDS) development of albuminuria or renal impairment was associated with retinopathy.24 A large American study of type 2 diabetes patients25 reported that albuminuria had strong independent associations with retinopathy, neuropathy and cardiovascular disease. Albuminuria may be more than an indicator of renal disease in type 2 diabetes patients and may reflect a state of generalised vascular damage occurring throughout the body. In this context it is interesting to note that, in our study, albuminuria but not raised creatinine was significantly associated with STDR in multivariate analysis.
Our study demonstrated a significantly higher prevalence of STDR in patients with type 2 diabetes using insulin compared with those not using insulin, as has been previously reported.4 6 19 Duration of diabetes was significantly associated with STDR in univariate but not multivariate analysis. Many population-based prevalence studies support a link between duration of diabetes and retinopathy in both type 14 and type 2 diabetes,4 6 including one study from South Africa.11 Factors that may affect our analysis include variable access to healthcare, and therefore delayed diagnosis of diabetes, and the possibility of faster progression of retinopathy in some patients affected by co-pathologies including anaemia.
The UKPDS showed that incidence and progression of retinopathy was reduced by intensive blood glucose control8 and tight blood pressure control.9 In our study STDR was associated with elevated systolic blood pressure in univariate but not multivariate analysis and was not associated with fasting glucose or HbA1c. STDR in patients with type 1 diabetes was associated with neuropathy. This finding is in accordance with the Diabetes Control and Complications Trial (DCCT), which demonstrated that intensive glycaemic control in type 1 diabetes subjects reduced progression of retinopathy, neuropathy and nephropathy.7 The small number of type 1 diabetes subjects in our study limits further analysis.
In our study STDR was not associated with HIV. Although rapid progression of DR in HIV-positive patients has been reported,26 no large studies have examined this relationship. There are many potential interactions. It is well established that HIV is associated with the development of a metabolic syndrome that can result in impaired glucose tolerance and diabetes.27 The non-nucleoside reverse transcriptase inhibitors, stavudine and lamivudine, which are among the first line antiretroviral treatments used in Malawi, are associated with development of diabetes.28 Complications of HIV may mimic diabetic microvascular disease. This is a potential confounding factor in our study in which 13.7% of patients were HIV positive.
Of patients in this study 8.5% were visually impaired and 3.6% severely visually impaired or blind. Vision worse than 6/18 was significantly associated with STDR, presence of clinically significant cataract in both eyes and increasing age. Therefore, among subjects in this study both DR and cataract appear to contribute to the burden of visual impairment. Few studies have reported VA data in southern African diabetic populations. Mash et al12 reported in 400 diabetes patients a prevalence of visual impairment of 28.8% and severe visual impairment or blindness of 11.4%—higher than in our study. The age of subjects in their study was not reported.
We recognise the limitations of our study. First, the study was conducted in a single centre based at a central hospital. Patients attending the clinic were mostly urban dwellers and so our results should be generalised to other settings only with caution. It is possible that our data underestimate retinopathy and hyperglycaemia as people in rural areas have less access to healthcare. Patients who do not comply with clinic attendance may be the same patients who do not comply with other treatments for diabetes. Those with established complications may be more likely to participate in research studies. Another potential bias is that the service is seen by many as a clinic for collecting medication, and so patients with diet-controlled diabetes are almost certainly under-represented. Despite these limitations, we believe that the size of our study and the degree of confidence around our findings render them useful.
In conclusion, our data provide an estimate of the prevalence of DR in patients attending a southern African diabetes clinic in which no publically funded service for detection and treatment of retinopathy is currently available. The prevalence of diabetes in Africa is increasing and there is an urgent need for service provision. This study provides information that can be used by healthcare providers in planning the introduction and development of diabetes services in the region. It is in comparison with data such as this that the efficacy and cost-effectiveness of future interventions can be assessed.
The authors acknowledge the help of Helen Dzamalala and Daniel Chimbayo in data collection (both affiliated to the College of Medicine, University of Malawi, Blantyre, Malawi).
Competing interests None.
Patient consent Obtained.
Ethics approval Ethics approval was obtained from the Malawi College of Medicine Research and Ethics Committee.
Provenance and peer review Not commissioned; externally peer reviewed.