ABSTRACT

Background: As part of a project to improve diabetes carein Fiji, we assessed the magnitude of problems posed bydiabetic retinopathy in that country and compared thefindings with those from an Australian diabetes centre.Therelationship between diabetic retinopathy and nephropathywas also examined in a subset of patients.

Methods: A medical team from Australia screened a total of446 type 2 diabetic patients (ethnicity: Fijian/Indian 16/84%)for diabetic retinopathy in five towns from the WesternDivision of Viti Levu, Fiji. The findings were compared withdata obtained from 1659 type 2 diabetic patients who hadattended an Australian diabetes centre (ethnicity: Indian/Anglo-Celtic 12/88%). In both cohorts, retinopathy wasassessed by direct fundoscopy and a spot urine sample wascollected for determination of albuminuria (defined as aconcentration > 50 mg/L).

Results: The prevalence of diabetic retinopathy increasedlinearly with duration of diabetes. It was higher in Fiji, evenwhen cases from the same ethnicity (i.e. Indians) and dura-tion were compared (P < 0.05). Extrapolation of the datapoints suggests a delay in the diagnosis of diabetes in Fiji.Of those patients with retinopathy in Fiji, more than halfhad moderate to severe non-proliferative diabetic retino-pathy or proliferative diabetic retinopathy, significantlyhigher than patients in the Australian cohort (χ2 = 29.2;P < 0.0001). Retinopathy was not a predictor of albuminuriain Fijian Indians (χ2 = 0.4; P = 0.5). In contrast, AustralianIndians with retinopathy had significantly more albuminuria(χ2 = 10.2; P = 0.001).

Conclusions: Severe diabetic retinopathy is common inboth ethnic groups in Fiji. A delay in the diagnosis of

diabetes as well as poor glycaemic control are possiblefactors. The availability of laser therapy is important toprevent loss of vision, but it is also essential that appro-priate training of health professionals is integrated with aprogramme of diabetic complication screening to supportthis form of therapy.

Key words: albuminuria, complications, health care, nephro-pathy, retinopathy.

INTRODUCTION

Diabetic retinopathy is the leading cause of blindness inpeople up to 65 years of age.1 As visual loss is preventablewith timely laser therapy, early detection and treatment areimportant aspects of public health. In Australia, currenthealth guidelines recommend that patients are routinelyscreened for diabetic complications. In developing countriessuch as Fiji, facilities for the detection and treatment ofdiabetic complications are less readily available.

In 1996, the Diabetes Centre of Royal Prince AlfredHospital (RPAH) was sponsored by the Australian Agencyfor International Development (AusAID) in a programmethat aimed to assist the prevention and management ofdiabetes in Fiji. As part of this project, a survey was con-ducted to assess the magnitude of problems posed bydiabetic retinopathy. To evaluate whether the presence andseverity of diabetic retinopathy differed in Fiji and Australia,the Fijian data were also compared with those obtained frompatients attending the RPAH. As retinopathy is a predictorof other diabetic complications, we also examined therelationship between diabetic retinopathy and albuminuria(the most sensitive index of nephropathy) in Fijian andRPAH Indian patients.

Australian and New Zealand Journal of Ophthalmology (1999) 27, 9–13

Original Article

Diabetic retinopathy and nephropathy in Fiji: Comparison withdata from an Australian diabetes centreBelinda Brooks, RN,1 Robert Chong, MB, BS,1 Ivan Ho, MB, BS,1 Fiona Capstick, RN,1 Lynda Molyneaux,RN,1 Than T Oo, MD,4 Malcolm Tester, FRACO3 and Dennis Yue, FRACP,1,2

1Diabetes Centre, Royal Prince Alfred Hospital, 2Department of Medicine, The University of Sydney, Sydney, 3St Vincent’s Hospital,Lismore, New South Wales, Australia and 4Lautoka Hospital, Lautoka, Fiji

■ Correspondence: Professor DK Yue, Diabetes Centre, Royal Prince Alfred Hospital, Camperdown, NSW 2050, Australia. Email: <dice@diab.rpa.cs.nsw.gov.au>

METHODS

Fijian cohort

A medical team from Australia, consisting of one ophthal-mologist, two nurses and two final year medical students,visited five towns in the Western Division of Fiji in Viti Levuand examined a total of 446 type 2 diabetic patients (Lautokan = 68; Ba n = 103; Sigatoka n = 88; Raki-Raki n = 92; Tavuan = 95), consisting of 16% Fijians and 84% Indians. In43 patients, neither fundus could be visualized due to thepresence of severe cataracts, leaving 403 patients (16%Fijians and 84% Indians) available for analysis. Patients weremade aware of the visit of the diabetic eye team through theirdoctors and a combination of radio announcements, news-paper advertisements and word of mouth. Thus, some patientswere attenders at the hospital clinics, others were directlyfrom the local communities. The examination was conductedin either the local hospitals or health care centres. Due tolanguage difficulties and the number of patients involved, itwas not possible to assess visual acuity. The retina was exam-ined by direct fundoscopy through pupils dilated with 1%tropicamide (Chauvin Pharmaceuticals, Essex, UK). Thevisiting ophthalmologist (MT) examined all patients and, inthe majority of cases, patients were also seen by the localophthalmologist. The severity of diabetic retinopathy wasgraded according to the worst affected eye and was classifiedas either normal, minimal-to-mild non-proliferative diabeticretinopathy (NPDR), moderate-to-severe NPDR or prolifer-ative diabetic retinopathy.2

In a subset of Indian patients (90 consecutive patients withretinopathy and 71 without retinopathy), spot urine albuminconcentration was determined using Micral-Test II (BoehringerMannheim, Mannheim, Germany). For the purpose of analy-sis, albuminuria was defined as a concentration > 50 mg/L.Sitting blood pressure (BP) and antihypertensive treatments ofthese patients were also recorded.

RPAH cohort

In order to compare complication status of patientsscreened in Fiji with those in Australia, the retinopathy

and albuminuria status of Indian (n = 192) and Anglo-Celtic(n = 1467) type 2 diabetic patients when they first attendedRPAH Diabetes Centre were retrieved from a computerizeddatabase. There were insufficient numbers of Fijian patientsin this database to allow meaningful analysis. Patientsattending the RPAH Diabetes Centre are community basedand are referred by their general practitioners specificallyfor diabetic complications assessment. All patients wereseen by one endocrinologist (DY) and were routinelychecked for the presence of diabetic retinopathy, albumin-uria, hypertension and its therapy. Direct fundoscopy wasperformed by the endocrinologist. Diabetic retinopathywas graded according to the same criteria used for the Fijiancohort. The validity of the retinal findings of the endo-crinologist has been verified against that of an ophthalmol-ogist. The results showed that, in comparison with theophthalmologist, the endocrinologist had a sensitivity of100% and a specificity of 80% and a positive predictivevalue of 93% for the detection of any retinopathy. Agree-ment between the two was also measured and there wasexcellent agreement (κ = 0.85; P < 0.0001). Albuminuria ofa spot urine sample was measured by radioimmunoassayand, again, was considered to be present at a concentration> 50 mg/L.

Statistical methods

The demographic parameters shown in Table 1 areexpressed as median and interquartile range. Kruskal–Wallis test was used to compare age, duration of diabetesand age at diagnosis between Fijian, Fijian Indians, RPAHIndians and RPAH Anglo-Celtics. Multiple comparison testswere performed using the Kruskal–Wallis Z-test. The rela-tive risk of albuminuria associated with retinopathy wascalculated using logistic regression and results are expressedas odds ratio (OR) and 95% confidence interval (CI).Categorical data were analysed using the Chi-squared test.Linear regression was used to examine the relationshipbetween duration and prevalence of retinopathy in thevarious ethnic groups studied. Statistical significance wasaccepted at P < 0.05.

10 Brooks et al.

Table 1. Profile of non-insulin-dependent diabetes mellitus patients grouped by ethnicity

Data show the odds ratio, with 95% confidence intervals given in parentheses. *Excludes patients fundus not examined.†P < 0.05 compared with Royal Prince Alfred Hospital (RPAH) Indians; ‡P < 0.05 compared with RPAH Anglo-Celtics.

Ethnicity Age (years) Duration of diabetes (years) Age at diagnosis (years)

Fijians (n = 64)* 57.5 (48.5–63.5)†‡ 5.0 (3.0–10.0)†‡ 49.5 (42.0–58.0)†‡

Fijian Indians (n = 339)* 56.0 (50.0–61.0)†‡ 9.0 (4.0–14.0)†‡ 47.0 (40.0–53.0)‡

RPAH Indians (n = 192) 50.2 (41.4–59.4)‡ 4.3 (0.6–8.9) 43.4 (36.9–51.6)‡

RPAH Anglo-Celtics (n = 1467) 59.8 (51.3–67.9) 2.9 (0.3–8.5) 54.0 (45.1–62.4)

RESULTS

Demographic profiles of the patients, grouped according toethnicity and country of origin, are shown in Table 1. TheFijian Indian patients are older and have had diabetes forlonger than their RPAH counterparts. The overall preva-lence of diabetic retinopathy was significantly higher inboth Fijians and Fijian Indians than in patients attendingRPAH (Fig. 1). Of those patients with retinopathy in theFijian cohort, more than half have either moderate-to-severeNPDR or proliferative diabetic retinopathy, significantly

higher than the RPAH population (χ2 = 29.2; P < 0.0001;Fig. 2). The prevalence of retinopathy by duration of dia-betes, for each of the ethnic groups, is shown in Fig. 3. Thisincreases linearly with longer duration of diabetes in allgroups. For any given duration of diabetes, the proportion ofpatients with retinopathy remains higher in the Fijian cohortcompared with the RPAH cohort. Multiple regressionanalysis showed 78% of the variance for retinopathy wasaccounted for by duration of diabetes for the Fijians, 92%for Fijian Indians, 86% for RPAH Indians and 99% for RPAHAnglo-Celtics. Using the technique of Harris et al.,3 byextrapolating the lines of best fit for each group to the pointat which retinopathy prevalence is 0, we can calculate thatthere is an approximate delay in the diagnosis of diabetes of9 years for Fijians, 11 years for Fijian Indians, 2 years forRPAH Indians and 6 months for RPAH Anglo-Celtics.

In Fijian Indian patients, albuminuria was present in 59%of patients with retinopathy and in 49% of those withoutretinopathy and retinopathy was not a predictor of albumin-uria (χ2 = 0.4; P = 0.5). In contrast, for RPAH Indianpatients, albuminuria was present in 40 and 13% of patientswith and without retinopathy, respectively (χ2 = 10.2;P = 0.001). Retinopathy was a predictor of albuminuria inthis situation (Table 2). The median BP of the Fijian Indianpatients was 140/90 mmHg and, overall, 50% were on anti-hypertensive therapy. The RPAH Indians had a significantlylower median BP of 128/80 mmHg compared with the FijianIndians (Z = 4.9, P < 0.0001 and Z = 3.5, P < 0.0004 forsystolic and diastolic BP, respectively). In addition, only30% of RPAH Indians were taking antihypertensive treat-ment (χ2 = 11.4; P = 0.0008).

Diabetic retinopathy and nephropathy in Fiji 11

Figure 1. Prevalence of diabetic retinopathy in Fiji and Australiagrouped by ethnicity. (h), Fijians; (j), Fijian Indians; ( ), RoyalPrince Alfred Hospital (RPAH) Indians; ( ), RPAH Anglo-Celtics.*P < 0.0001, ‡P < 0.001 compared with RPAH Anglo-Celtics;†P < 0.0001 compared with RPAH Indians.

Figure 2. Classification of dia-betic retinopathy in (a) Fijian and(b) Australian cohorts. The piechart shows the frequency of noretinopathy ( ) and retinopathy(j) in the (a) Fijian (47.7 and52.6%, respectively) and (b) Aus-tralian (86.7 and 13.3%, respec-tively) cohorts. The bar chart onthe right indicates the classifi-cation of retinopathy as prolifera-tive (h), mild non-proliferativediabetic retinopathy (NPDR; )or severe NPDR ( ) in the (a)Fijian (9.9, 42.9 and 47.2%,respectively) and (b) Australian(7.6, 69.0 and 23.3%, respec-tively) cohorts.

DISCUSSION

Retinopathy is a serious and yet the most treatable diabeticcomplication provided it is detected early and facilities fortreatment are available. The present study clearly shows thatretinopathy is common in Fiji. Moreover, vision-threateningretinopathy requiring laser treatment constitutes more thanhalf the detected cases in Fiji. These findings are signifi-cantly worse than those from Australian patients, even whencases from the same ethnicity (i.e. Indians) were examined.While it is possible that the two Indian populationsorginated from different parts of India, these types of demo-graphic data were not collected.

Duration of diabetes and glycaemic control are the twofactors known to be strongly associated with the develop-ment of diabetic retinopathy. Initial examination of our datareveals that the higher prevalence of diabetic retinopathyamong Fijian patients remains evident, even when adjustedfor duration of diabetes. This would suggest that worseglycaemic control in the Fijian diabetic population is thebasis of our observation. However, the data can also beinterpreted in another way. The relationship between thepresence of retinopathy and duration of diabetes is remark-

ably linear in every ethnic group examined. If we use themethod of Harris et al.,3 we can extrapolate the regressionline obtained from direct patient observations to allow us toderive the theoretical time point where there would be azero prevalence of retinopathy. This manoeuvre shows that,on average, the diagnosis of diabetes in Fiji was delayed byapproximately 10 years, whereas in the RPAH cohort thedelay was negligible. Adjustment of this delay in diagnosisby shifting the Fijian lines to the right would obliterate thedifference between the two countries. In this interpretation,delay in diagnosis (hence, a falsely low duration of diabetes)rather than worse glycaemic control would be the maincause for the difference in the presence of retinopathy thatwe have observed. This does not imply that poor glycaemiccontrol is playing no part in our data. It could be a factorexplaining the higher percentage of vision-threateningretinopathy in the Fijian cohort.

Although differing sensitivities in the detection of retino-pathy between the two countries cannot be completely dis-counted, because fundal examination was conducted by twodifferent individuals, we feel that it is unlikely to be a majorfactor in our results. The endocrinologist involved at theRPAH had examined 10 000 fundi in the past 12 years andhad demonstrated a low false negative rate in detecting andassessing the severity of retinopathy when checked againstan ophthalmologist. Obviously, retinal photography wouldhave provided better documentation, but this was not logis-tically possible. It is also evident that our study was notpopulation based, such as the studies of Klein et al.4,5 andMitchell et al.,6 and there may have been some degree of selfselection of patients when they learnt that eye examinationwould be available. The fact that a number of patients in Fijihad bilateral cataract and could not be assessed could alsohave some impact on the prevalence of retinopathy.

It is interesting that although many studies, including ourfindings in RPAH Indians, have found retinopathy and albu-minuria to be associated, this relationship was not demon-strable in the Fijian Indians. It is known that certainpopulations, the Australian Aborigines among them, have ahigh prevalence of kidney disease independent of diabetesand this may explain the loss of association between thesetwo complications in Fiji. The higher prevalence of hyper-tension and its treatment in Fiji supports this possibility, buta more detailed study is required to clarify this point.

Naturally, in our endeavour to combat diabetic retino-pathy, early detection of diabetes and better metaboliccontrol are essential. In addition, one clear conclusion thatcannot be disputed is that more resources need to be investedin the detection and treatment of diabetic retinopathy in Fiji.In addition to the common finding of retinopathy, manypatients with severe cataract were also found; in all likelihooddiabetes is also playing a role. Many cases with proliferative

12 Brooks et al.

Figure 3. Relationship between the duration and prevalence ofdiabetic retinopathy in different ethnic groups in Fiji and Australia.(s), Fijians; (d), Fijian Indians; (,), Royal Prince Alfred Hospital(RPAH) Indians; (.), RPAH Anglo-Celtics.

Table 2. Risk of albuminuria for patients with retinopathy

OR, odds ratio; CI, confidence interval.

Ethnicity OR 95% CI χ2 P value

Fijian Indians 1.2 0.6–2.5 0.4 0.5RPAH Indians 4.4 1.8–10.9 10.0 0.001

retinopathy had progressed to the end-stage of vitreoushaemorrhage, retinal detachment and rubeosis iridis, many ofwhich may have been prevented with early retinal laser inter-vention. In order to reduce diabetic retinopathy, patientsneed to have easy access to the health care system, regularsupplies of medications and patient education facilities toimprove diabetes self management. Thus far, facilities for eyescreening in Fiji are limited and no laser photocoagulationfacility exists in the Fijian public health system, although twoprojects sponsored by prominent service clubs in Australiaand New Zealand are currently seeking funds to implementsuch services. Even if laser treatment was to become avail-able, it is essential that promotion of public awareness andappropriate training of health professionals are integratedwith a programme of diabetic complication screening tosupport this form of therapy.

ACKNOWLEDGEMENTS

This work was sponsored in part by an AusAID grant. Theauthors thank the Ministry of Health, Fiji, and the NationalDiabetes Foundation, Fiji, for their co-operation and assis-tance in organizing this study. We are indebted to the

doctors and nurses of Lautoka Hospital, Sigatoka Hospital,Ba Hospital, Tavua Hospital and Raki-Raki Hospital for theirassistance and support.

REFERENCES

1. National Society to Prevent Blindness. Vision Problems in theUnited States. New York: National Society to Prevent Blindness,1980.

2. Management of Diabetic Retinopathy. Clinical Practice Guidelines.Canberra: National Health and Medical Research Council,1997.

3. Harris MI, Klein R, Welborn TA, Knuiman MW. Onset ofNIDDM occurs at least 4–7 yr before clinical diagnosis.Diabetes Care 1992; 15: 815–19.

4. Klein R, Klein BEK, Moss SE, Davis MD, DeMets DL. TheWisconsin Epidemiologic Study of Diabetic Retinopathy.Prevalence and risk of diabetic retinopathy when age at diag-nosis is less than 30 years. Arch. Ophthalmol. 1984; 102: 520–6.

5. Klein R, Klein BEK, Moss SE, Davis MD, DeMets DL. TheWisconsin Epidemiologic Study of Diabetic Retinopathy. III.Prevalence and risk of diabetic retinopathy when age at diag-nosis is 30 or more years. Arch. Ophthalmol. 1984; 102: 527–32.

6. Mitchell P, Smith W, Wang JJ, Attebo K. Prevalence ofdiabetic retinopathy in an older Australian community: TheBlue Mountains Eye Study. Ophthalmology 1998; 105: 406–11.

Diabetic retinopathy and nephropathy in Fiji 13