Methods

Sample population and plan

The population of the seven districts in the study (including Chitwan) was estimated at 3.36 million as of 2006, out of which 13% (0.43 million) were aged 50 years and older with 50% women.3 We utilised data from the 2001 national population census as the frame for the sampling design.

Sample size

The sample size was calculated to estimate the prevalence of blindness (defined as presenting better-eye visual acuity of <6/60) for people aged 50 years and older of 5%.3 A sample size of 5000 was based on a 95% CI, 15% error-bound precision, 85% examination response rate, cluster design effects of 2.00 and rounding up to the nearest 1000 people.

Sampling methodology

We combined small villages and divided larger villages to create sampling clusters with nearly 175 (108 to 216) people ≥50 years (total population between 850 and 1700) in each cluster. We used simple random sampling without replacement (based on the 2001 population census estimate of 12.75% population aged 50 years and older and a total survey population of 39 216 people) to select 32 clusters for the study.

Study team

The study included two enumeration teams and one clinical team. Each enumeration team comprised an enumerator-supervisor, enumerator, map-maker and a helper or a village volunteer. The clinical team consisted of an ophthalmologist and two ophthalmic assistants. Local volunteers were enlisted to assist with the clinical examination. Both enumeration and clinical teams had at least one female member.

Field procedures

Fieldwork took place over 20 weeks during August to December 2006.

Enumeration

All individuals aged 50 years and above in all households in the selected clusters were enumerated through door-to-door visits. Each household was visited at least three times on consecutive days before a member was designated as unavailable. Suitable sites, ideally equipped with electricity, ambient light and adequate examination and waiting areas were selected within a 1 h walking distance of selected households.

Clinical examination

Ophthalmic assistants performed a visual-acuity assessment at presentation with and without spectacles using a back-illuminated Early Treatment for Diabetic Retinopathy Study (ETDRS) E chart at a distance of 4 m. Those unable to read the ETDRS chart at 4 m were asked to read the chart at 1 m, then finger counting at 1 m, hand movements and light perception. Where vision could not be tested, the ophthalmic assistant used their judgement to make a subjective determination and record “believed blind,” “believed not blind” or “undetermined.”

All people presenting with a visual acuity of less than 6/18 and those operated for cataract surgery in either eye were sent for refraction, while others proceeded directly to the ophthalmologist for detailed eye examination.

Ophthalmologist's examination

The ophthalmologist used a slit lamp, torch light and direct ophthalmoscope for the basic eye examination and to determine the type of cataract surgery and surgical complications. The depth of the anterior chamber was graded against the corneal thickness as “normal, shallow, deep or undetermined.” The ophthalmologist performed a fundus examination using direct ophthalmoscopy in a dimly lit room. People with dense cataracts or requiring further examination were referred to the Lumbini Eye Institute for indirect ophthalmoscopy. Intraocular pressure was measured using an applanation tonometer.

The examining ophthalmologist determined the principal cause of blindness or low vision for each eye and recorded the appropriate code as per a prespecified list. Only one condition was marked as the principal cause, and where there were contributory causes, provision was available to record two main causes.

Research ethics and field conduct

Verbal informed consent was obtained from participants. The examination protocol was cleared by the WHO Secretariat Committee on Research Involving Human Subjects. The project was approved by the Ethical Review Committees of Nepal Netra Jyoti Sangh (National Society for Prevention of Blindness) and the Lumbini Eye Institute. Patient confidentiality was maintained as per protocol.

Quality assurance

During the prepilot and pilot stages, inter- and intraobserver data were collected, and analysed by the Technical Advisory Committee. During the study period, quality assurance was tested in five randomly selected clusters by duplicating ophthalmic assistant examination for people who presented with VA <6/18 and in 10% of people with normal vision. The interobserver agreement for presenting VA between vision categories was 98% (unweighted kappa 0.92) and 91.2% (unweighted kappa 0.89) for best-corrected VA.

Data analysis

Five vision categories were defined for analysis and reporting: (1) normal or near normal vision, 6/18 or better in both eyes; (2) visual impairment, unilateral or bilateral visual impairment, <6/18 to 6/60 in the worse eye and >6/60 in the better eye; (3) unilateral blindness, <6/60 in the worse eye and >6/60 in the better eye; (4) moderate bilateral blindness, <6/60 in worse eye and <6/60 to >3/60 in the better eye; (5) severe bilateral blindness, <3/60 in both eyes. Estimates (with 95% CIs) for impairment and blindness prevalence were calculated along with that attributed specifically to cataract. The prevalence of blindness, cataract blindness and cataract surgery was estimated, and potential associations with age, sex, literacy and geographic location were explored using multiple logistic regression.

The cataract blindness burden was defined as the sum of those people already operated on for cataract in both eyes and unoperated cataract blind. It was not possible to obtain the preoperative vision status on an already operated eye, so we assumed that both eyes were blind preoperatively if both eyes were operated for cataract, or if one eye was operated, and the other eye was blind at the time of our examination. Surgical coverage was calculated as the number of bilaterally blind cataract cases operated on divided by the number who could have been operated on. The denominator includes the already operated bilateral blind (the numerator) plus the unoperated bilaterally blind with cataract being the principal cause of blindness in at least one eye.

CIs for prevalence estimates and odds rations were calculated. We considered a p value <0.05 as significant. Missing values were assumed to be similar in distribution to available data and ignored during the analysis.

Results

Of the 5916 people enumerated, 5138 (87%) were examined, ranging from 68% to 96% among sampling clusters. Of the 778 people not examined, 80% were working or visiting relatives outside the village and were presumed not to have significant visual impairment. Only seven people refused eye examination.

No significant demographic differences were found between the enumerated and examined populations (table 1). Women constituted 52% and 53% of the total enumerated and examined population, respectively, compared with an estimated 50% women in the population in the Lumbini and Narayani zones.3

Agriculture (46%) was the only significant income-earning occupation, and women provided essentially all the household work (99%).

Table 2 shows the prevalence of visual impairment and blindness based on presenting visual acuity (with spectacles if presenting with them) and best-corrected visual acuity. Severe (<3/60) and moderate (<6/60) bilateral blindness in the best eye were estimated as 2.3% each, and visual impairment (<6/18) in the worst eye as 18.9%.

Table 3 provides the prevalence of presenting visual impairment and blindness in people, by age, sex, literacy and geographical location. Almost one-third of the population (32.3%) have a visual problem, with 19% having visual impairment in at least one eye and 8.8% unilateral blindness (<6/60). There was a slightly higher prevalence of visual impairment and blindness in men compared with women (19.9% vs 18.1%). However, the distribution of presenting vision between men and women was not statistically significantly different across any of the five visual acuity categories. People aged 60–69 years (adjusted OR 3.2, 95% CI 2.2 to 4.6) and people aged 70 years and older (adjusted OR 6.1, 95% CI 4.1 to 9.1) had a statistically significantly greater prevalence of bilateral blindness compared with people aged 50–59 years. There was a significantly higher prevalence of blindness among people considered illiterate versus literate (adjusted OR 2.9, 95% CI 1.6 to 5.1) and living in the plain versus the hill districts (adjusted OR 0.6, 95% CI 0.4 to 0.9).

Cataract (49%), refractive error (31%), corneal opacity (3.8%) and retinal disorders (4.0%) were the leading causes of unilateral and bilateral blindness in eyes (table 4). Of people with bilateral blindness (<6/60), 48% have cataract in both eyes, and 31% have bilateral refractive error.

Table 5 provides estimates of cataract blindness and surgical coverage by age, sex, literacy and residence. In the multivariate analysis, cataract blindness burden was significantly higher with age (OR 9.8, 95% CI 6.7 to 14.4) and illiteracy (OR 1.8, 95% CI 1.2 to 2.6). The cataract blindness burden was significantly higher in the plain than the hill districts (OR 1.51, 95% CI 1.21 to 1.88).

Discussion

The survey findings suggest that blindness prevalence may have decreased since the 1995 survey.2 However, during this period, the population aged 50 and older in the Zone and Chitwan District increased from 390 000 to 435 000 with a corresponding increase in life expectancy from 54 to 63 years. Any rigorous comparison would require calculation of age and sex standardised rates to account for the ageing population and possible changes in sex distribution.

Using point estimates, the 2006 survey found a reduced blindness prevalence (4.6%, 95% CI 3.4 to 5.8; presenting vision <6/60) in people aged 50 years and older, compared with people age 45 and older (5.3%) in the 1995 survey involving the Lumbini Zone.2 A blindness estimate for the comparable population (≥age 50) from the 1995 survey almost certainly lies outside the 95% CI for the 2006 survey. Using the CIs from 2006, the severe blindness prevalence estimate was also significantly reduced in 2006 (2.3%, 95% CI 1.7 to 2.8; presenting vision <3/60) compared with people aged 45 and older in 1995 (3.0%)2 and 1981 (3.8%).1

This 2006 Lumbini survey estimates a blindness (<6/60) prevalence of 4.3% for women and 5% for men, respectively. This represents a reversal of the 1995 ratio which favoured men (4.6%) over women (6.1%). The reversal was almost certainly due to increased cataract surgical coverage in women (70.8%) versus men (61.7%) compared with the 1995 estimates (women 40.6%, men 44.2%). Reducing gender inequity has been an important outreach programme emphasis at LEI since 2001. Beginning in the late 1990s, LEI trained female community health volunteers to identify and support women both to attend its expanding network of regular screening camps and to undergo surgery either in the camp or more often back at LEI or Bhartpur Hospital in Chitwan.

This 2006 Lumbini survey estimates a significantly lower blindness (<6/60) prevalence in the hill (3.3%) versus plain region (5.8%) (adjusted OR 0.6, 95% CI 0.4 to 0.9) despite very similar demographics (age and sex). The blindness prevalence in the plain area is consistent with estimates of Indian populations (8.5%) which they resemble ethnically.4 However, although the hill population consists of more of an “Asian” population, its prevalence is much lower than blindness (<6/60) estimates for people over the age of 50 (10.9%) in the Tibet Autonomous Region of China,5 for example.

The higher prevalence of blindness in the plain area is most likely due to a larger backlog of unoperated cataract cases. The cataract blindness burden (sum of unoperated and operated, presumed blind) was found to be higher in the plain (8.9%) than in the hill (6.1%) areas (OR 1.51, 95% CI 1.21 to 1.88) and the cataract surgical coverage was almost the same. LEI achieves the cataract surgical coverage in the hill districts through an eye hospital in Tansen, the only hospital in the hill districts of Nepal, and a network of regular screening visits and surgical eye camps conducted by the resident ophthalmologist.

The significantly lower blindness prevalence estimate in the hill versus plain region is similar to the 1981 Nepal National Blindness survey.1 They also attributed the lower prevalence of blindness in the hill Districts primarily due to the lower prevalence of cataract compared with the central plain (Tarai) areas. Current (2009) population-based studies in Nepal plan a further examination of these geographical patterns.

The current Lumbini blindness (<6/60) prevalence estimate is comparable with people age ≥50 years in the Satkhira District in Bangladesh (presenting vision 4.5%)6 The Lumbini blindness prevalence (presenting vision <3/60) estimate of 2.3% is lower than the 2002 estimate of 3.4% for the South East Asian Region7 and lower than the Pakistan National Study in people aged ≥50 which estimated a blindness (<3/60) prevalence of 7%.8

A survey conducted in the adjacent Gandaki Zone of Nepal in 20029 found a lower blindness (<6/60 presenting vision) prevalence for people age 45 and older (2.6%) compared with the 2006 Lumbini study (4.6%). However, the Gandaki study only sampled patients from the three Districts closest to the base hospital where programme impact was likely greatest, and Gandaki consists of only hill regions similar to the hill Districts of Lumbini, where the prevalence estimate was similar (3.3%).

Cataract remains as the principle cause of blindness in Lumbini Zone (48% of eyes) but is substantially lower than the 1995 estimates (78%)2 and the current national estimate of 70%.10 This is consistent with the cataract surgical coverage of 66% compared with the Nepal national estimate of less than 50%.10 The current cataract surgical coverage (66% <6/60) is also much higher than in 1995 in the Lumbini Zone (42%).2

The cataract surgical coverage over the 10-year period not only increased from 42% to 66%, but also showed a more equitable distribution of service delivery by sex and geographic location. The surgical coverage became more rather than less common among women (71%) versus men (62%) and similar in the plain (68%) and more remote hill Districts (64%).

From 1995 to 2006, the number of cataract operations in Lumbini Zone increased from approximately 4000 (3300 LEI +700 private and conducted elsewhere) to 6700 (5700 LEI); a cataract surgical rate of 2666 per million. Using the survey estimate of unoperated cataract blindness (<6/60) as 2.3% (50% of total blindness) there was a backlog of approximately 8300 cases in 2006.

For programme planning, in keeping with IAPB planning in India,11 LEI should also consider both the incidence of cataract and operations on eyes with vision >6/60, second eyes, and unilaterally blind people. LEI should expect an annual incidence of 0.45% (20% of the prevalence)12 or 1600 people needing surgery in at least one eye. In addition, LEI estimates that, of its current operations, approximately 10% of operated eyes (670) have vision better than 6/60, and at least another 10% (670) are second eyes or unilaterally blind people. Therefore, approximately 50% of the annual cataract operations (3000) reduce the backlog of blind eyes (8700), and the other half deal with incident case and non-blind people.

Refractive error accounted for 31% of bilaterally and 12.9% of unilaterally blind people. Refractive error as the principal cause of blindness is greater than the Gandaki survey, 13.2% of bilateral and 10.2% unilateral blind.3 In the 1995 Lumbini survey, there where only 3.8% of eyes blind due to refractive error,2 and it was not a cause of blindness in the 1981 National Nepal Survey. The increased prevalence likely reflects an older population with more age-related cataract and no access to optometric services.

Conclusions

In the past 10 years, blindness prevalence, particularly due to cataract, has decreased, and cataract surgical coverage has increased in the Lumbini Zone and Chitwan District. Cataract and other surgical services are equitably distributed by age and sex throughout all Districts in the service area. Blindness due to refractive error has increased tenfold in the past 10 years. With 79% of bilateral blindness attributable to cataract and refractive error, there is substantial scope for further reduction in burden of blindness in this population. The current cataract surgical rate will gradually decrease blindness due to cataract in the zone.