Pakistan, the sixth most populous country in the world,¹ is a developing country situated in the World Health Organization’s (WHO) Eastern Mediterranean Region. The country ranks 135 in the United Nations Human Development Index,² and a recent report has suggested that the country is facing significant hardship; a declining growth in gross domestic product (GDP) and a near doubling of the proportion of the population living below the poverty line between 1987 and 2003.³ The geography and climate of Pakistan are extremely diverse; the eastern and southern parts are dominated by the Indus River and its tributaries, the northern parts by the snow-covered Himalayan mountain range. The country’s four provinces are Punjab, Sindh, North West Frontier Province (NWFP) and Balochistan.

The evidence base on national blindness and visual impairment in Pakistan prior to this survey is extremely limited. One study, estimating the main cause of blindness to be cataract (66.7%),⁴ led the National Committee for the Prevention of Blindness (NCPB) to develop a Five Year National Plan for the Prevention of Blindness (1994–1999) with a particular focus on large-scale expansion of cataract surgical services.

The aim of this second national survey (conducted between 2002 and 2004) was to apply more rigorous survey methodologies to produce accurate data. Details of the prevalence of blindness among adults (aged ⩾30 years) have been published⁵ and we now report on the causes of blindness and visual impairment, providing estimates of the magnitude of the major causes and exploring their demographic associations.

SUBJECTS AND METHODS

A detailed description of the methods used in the survey for sampling and training, and ocular examination protocols has already been published.⁶ A brief summary of the key methodological details is provided below.

Sample size

Based on an assumed prevalence of blindness of 1.8%, a random sampling error precision of 0.3%, a design effect of 2.0 and a 10% increase for non-response, the total sample size was calculated as 16 600.

Sampling strategy

Multi-stage stratified cluster random sampling, with probability proportional-to-size (PPS) procedures, was used. Enumeration, using the random walk method, was undertaken until the target number of adults was attained in each cluster. All enumerated individuals were asked to attend the survey station, set up in their community, for ophthalmic examination in the following days. Enumerated individuals who were unable to attend were examined in their home whenever possible. Three visits were made to homes before marking the subject as a non-responder.

Ethical and official government approval

Ethical approval was provided by the Pakistan Medical Research Council (PMRC) in March 2002. This study followed the tenets of the Declaration of Helsinki.

Definitions used in the ophthalmic examination

Visual impairment

The WHO categories of visual impairment were used in this study.⁷ Blindness (BL) was defined as a presenting visual acuity (ie, with glasses for distance if normally worn, or unaided) of <3/60 (<20/400 (logarithm of minimum angle of acuity (logMAR) >1.30) in the better eye. Severe visual impairment (SVI) was defined as <6/60 to ⩾3/60, and moderate visual impairment (MVI) as <6/18 to ⩾6/60. A “near normal” category was also included as <6/12 but ⩾6/18. Any person with an acuity of <6/12 in the better eye was regarded as visually impaired. As visual fields were only assessed on a subset of the sample, constricted visual fields were not included in the definition of blindness. The Snellen notation for visual acuity has been used herein for ease of comparison.

Unilateral SVI/BL

This was defined as a participant presenting with ⩾6/12 in one eye and <6/60 in the other eye.

Clinical examination

Distance visual acuities were measured in all subjects using a reduced logMAR tumbling “E” chart.⁸ All study participants had a basic eye examination and all also underwent autorefraction (Nikon Retinomax K-plus II Nikon, Tokyo, Japan). Individuals with <6/12 presenting visual acuity in one or both eyes (“red card holders”) were subject to a more detailed examination. All “red carders” had their visual acuity retested with the autorefraction result in a trial frame and all had a slit-lamp examination with a dilated posterior segment examination. All “red carders” presenting with <6/18 with a treatable condition were referred to the nearest eye care facility.

Identification of causes of visual loss

The survey ophthalmologist, epidemiologist and the three clinical ophthalmologists determined the cause(s) of visual loss, following the principles outlined in the WHO Prevention of Blindness Proforma (Version III). For each eye, all pathological findings were recorded at the time of examination. Any degree of improvement in visual acuity when retested with a refractive correction was deemed evidence of refractive error present in that eye. One main cause was then selected for each eye, the WHO recommendations stipulating that (1) if any pathology is secondary to another, the primary pathology should be selected (eg, if the pathology was band keratopathy secondary to uveitis, uveitis should be selected), and/or (2) conditions amenable to treatment or (3) which could have been prevented are preferentially selected over and above unavoidable causes. Following this, the main cause in the right eye or the main cause in the left eye was chosen to represent the principal cause for the individual. If the main causes in the right eye and the left eye differed, the principal disorder for the individual was selected as the one most readily treatable or, if not treatable, the one which was more amenable to prevention (eg, if the main causes were right eye cataract and left eye optic atrophy, cataract was selected as the principal cause). Refractive error was considered more amenable to treatment than cataract.⁹ If refractive error and cataract co-existed in the same eye, cataract was given as the main cause if refractive error correction did not improve the visual acuity ⩾6/18.

Statistical analysis

All data were double entered by two trained data processors. Conditions were subgrouped as preventable or treatable (ie, avoidable), or unavoidable. Cause-specific proportions of blindness and visual impairment were determined by age group, gender, province, location of household (rural/urban) and level of literacy. Univariate and age/gender-adjusted logistic regression modelling was used to explore associations of the major causes (eg, cataract vs all other causes) with demographic variables. Generalised estimating equations were used in the models to adjust for dependency in the data due to clustered sampling. Estimation of the cause-specific magnitude in Pakistan was calculated from age- and gender-standardised prevalence data (using the latest population data).¹ Extrapolations for the year 2020 were calculated using projected population estimates derived from the US census bureau.¹⁰

RESULTS

Study population

A total of 17 314 adults (⩾30 years) were enumerated, 16 507 (95.3%) of whom were examined and included in this study. Of the study sample, 53.1% were women, their mean age being significantly lower than that of males (45.9 vs 48.9 years, respectively, p<0.001). The demographics of responders and non-responders, and details of the prevalence of visual impairment have been described in detail elsewhere.⁵ A total of 4416 subjects (27%, 95% CI 26.1% to 27.4%) were identified with a presenting visual acuity <6/12. Of these, 561 subjects (crude prevalence 3.4%, 95% CI: 3.1% to 3.7%) presented blind.

Causes of bilateral blindness and visual impairment

Initially, all possible pathologies of a reduced visual acuity in eyes that presented with <6/12 vision were recorded by the examining ophthalmologist (total of 14 881 eyes). Refractive error and cataract were recorded as causes in 5463 (36.7%) and 5345 (35.9%) eyes, respectively. The next most common cause was central corneal opacity (912 eyes, 6.1%), uncorrected aphakia (430 eyes, 2.9%) and macular degeneration (418 eyes, 2.8%).

Data were then analysed using the principal cause as shown in table 1. Overall, an extremely high proportion of all categories of visual loss were due to conditions which could have been treated or prevented. A striking 85.4% of blindness was due to avoidable causes. Unoperated cataract together with uncorrected aphakia and posterior capsule opacification (PCO) accounted for 46.8, 78.1 and 63.7% of MVI, SVI and blindness, respectively. Under/uncorrected refractive errors accounted for 70.2% of visual loss in individuals with <6/12, but only 2.7% of blindness.

Amongst the 47 blind from “other” causes, posterior segment disorders dominated, including retinitis pigmentosa (11 subjects), vitreous haemorrhage (six subjects) and retinal detachment (two subjects).

Causes of unilateral SVI/BL

The main causes of unilateral reduced vision are presented in fig 1. Overall, more men were unilaterally blind than women (238 men, 187 women, p<0.001).

Demographic variation

We have previously reported that the prevalence of blindness increased almost exponentially with increasing age. The prevalence was also associated with female gender, rural dwelling and illiteracy.⁵

Among blind subjects, cataract was the main cause in all age groups. There were no persons blind as a result of glaucoma or uncorrected aphakia in the 30–39 year age group; however, among those aged 70 years and older, glaucoma and uncorrected aphakia accounted for 9 and 10%, respectively. PCO was not a cause of blindness in the younger age groups (ie, 30–59 years) but in older subjects it was a prominent treatable cause (6.3% in 60–69 year olds). In contrast amblyopia was a more common cause in younger subjects but was not reported as a cause in subjects ⩾60years. The highest proportion of phthsis/absent globe as a cause was found in the youngest age group (7.8%). The distribution in individuals with MVI is shown in fig 2.

The principal causes of MVI stratified by demographic variables are presented in table 2. In men, the principal cause was cataract (45.4%), whereas in women it was refractive error (45.4%). Provincial differences suggested that refractive error was a more common problem in NWFP (48.5%). Similarly refractive error was more common in urban settings (47.5%) (whereas in rural settings cataract dominated (45.3%)) and in literate subjects (59.3%). The proportion attributed to corneal opacity was highest in Balochistan (9.9%).

Association analysis

The age- and gender-adjusted association analyses of the principal cause in participants with SVI/BL are presented in table 3. The odds of refractive error as a cause compared with any other cause steadily decreased with age (p = 0.025), whereas the odds of cataract and aphakia increased with each decade (p = 0.023 and p = 0.025, respectively).

Estimate of the number of adults presenting with visual impairment in Pakistan by cause

If the acuity level of <6/60 is used to denote “operable cataract”, there are 904 000 (95% CI 736 000 to 1 107 000) adults requiring surgery. A further 173 000 individuals have uncorrected aphakia or PCO. A total of 1 390 000 adults have a presenting visual acuity of <6/60 in the better eye due to avoidable causes (table 4). Assuming the prevalence and patterns of causes remain unchanged, the figures for the year 2020 show that a total of 2 560 000 adults will have avoidable causes of SVI and blindness. The estimate for “operable cataract” is predicted to increase to almost 2 million by the year 2020.

Projections for the year 2020 are shown in table 4. Regarding adults with MVI (<6/18 to ⩾6/60, better eye), 2 140 000 would benefit from having their refractive error corrected, this number increasing to 4 320 000 adults in 2020.

DISCUSSION

The survey reported in this paper used a diagnostically rigorous methodology, similar to that used in the recent surveys in Bangladesh and India.^11,12 The response rate was very high (95.3%), minimising potential non-response bias. The use of the WHO criteria for coding causes of visual loss allows comparisons with other similarly coded surveys.

The refractive status of every individual was assessed by autorefraction, and, furthermore, subjects with a visual acuity of <6/12 in either eye had trial lens-corrected visual acuities measured. This allowed an accurate evaluation of refractive error as a cause of visual impairment, these data being important for planning refractive error services (a priority of the global initiative to eliminate avoidable blindness, VISION 2020: The Right to Sight¹³). The examination process also involved a dilated examination of all eyes with a visual acuity <6/12, allowing the detection of posterior segment disease, which has often been overlooked in the presence of cataract.¹⁴

For logistical reasons, fields were not performed on all subjects. Visual fields were only performed on a subset of the sample for normative data collection and on those who had optic disc changes that were suggestive of glaucoma, hence the decision not to use a definition of blindness that involved visual field construction. As fields were not performed on all subjects, constricted fields were not included in our definition of blindness. As the primary cause of a phthisical/absent globe could not always be determined, it is possible that some “unavoidable” cases could have been treatable or preventable (misclassification bias).

We found 14 881 eyes with visual acuity <6/12. When all pathological findings in each eye were analysed together, refractive error caused as much visual loss as cataract (36.7 and 35.6%, respectively).

Almost 75% of individuals who were blind had treatable causes, and >90% of subjects had treatable causes of visual impairment. The two most important treatable causes of blindness were unoperated cataract (or uncorrected aphakia, 8.6%; and PCO, 3.6%) and glaucoma (7.1%), others being refractive error (2.7%) and diabetic retinopathy (0.2%). In the 1990 study,⁴ cataract accounted for 66.7% of blindness, whereas the current survey found unoperated cataract accounting for 51.5%. It is not possible to compare data for cataract between these studies directly, as different age groups were used, but the observed reduction almost certainly represents a real reduction, given the large-scale increase in cataract surgical service delivery in Pakistan. However, the finding that despite this increase nearly 1 in 10 adults in Pakistan were visually impaired (<6/12) due to unoperated cataract highlights the importance of continued support of the NCPB for extending cataract surgical services in Pakistan.

In this survey, >12% of blindness was due to the sequelae of cataract surgery (ie, uncorrected aphakia 8.6% and PCO 3.6%). A survey conducted in rural NWFP identified uncorrected aphakia as the second most common cause of blindness.¹⁵ A hospital-based study in Lahore, Pakistan showed that only 50% of eyes among individuals returning for follow-up after cataract surgery had had intraocular lens (IOL) implantation.¹⁶ As low-cost, high-quality IOLs are now readily available, IOL surgery should be routine. Visual loss from PCO is certain to increase as cataract surgical rates increase, and YAG (yttrium–aluminium–garnet) lasers need to be made available for hospitals delivering high-volume cataract surgery, with training in their use, upkeep and repair made a priority.

After cataract and the sequelae of cataract surgery, glaucoma was the next most important cause of treatable blindness, accounting for 7.1%. This is lower than the 11% quoted for the WHO Eastern Mediterranean region, subregion D, which includes Pakistan,¹⁸ but much higher than in a similarly designed survey in Bangladesh (1.2%). A survey in India of adults aged ⩾50 years, which used a blindness definition of <6/60, showed that 5.8% of blindness was due to glaucoma.¹⁹ The earlier study in Pakistan estimated the number of people blind from glaucoma to be 80 000,⁴ which is similar to the estimate from the current survey (89 000). As cataract is becoming increasingly controlled in developing countries, strategies for the detection and treatment of glaucoma will need to increase in profile. In this survey, glaucoma blindness was highest in Balochistan, the most deprived province (89% of the rural population live in high deprivation districts)²⁰ with limited access to eye care services.

Diabetic retinopathy accounted for 0.2% of blindness. However, this is likely to underestimate the true burden of retinopathy in the population as diabetic patients are more likely to have cataract, which would be preferentially recorded as the cause of their visual loss, and vitreous haemorrhage (possibly from diabetic neovascularisation) was classified in the ‘other’ category. It is predicted that with rapid urbanisation, Pakistan will be among the five countries with the highest prevalence of diabetes by the year 2025.²¹ This is likely to alter the existing pattern of blindness and, in order to prevent a public health problem, preventive strategies need to be established.

In the 1990 Pakistan study, refractive error (including uncorrected aphakia) accounted for 11.4% of blindness,⁴ which is identical to the figure in the current survey, indicating that there has been minimal progress in addressing this highly treatable condition. Refractive error was the leading cause of MVI, particularly among the economically active working age group. Targeting this group, particularly with ready-made spectacles, should be made a priority as this would prove to be extremely cost effective. It must be recognised, however, that barriers exist—for example, the social implications of spectacle wear. In addition, many individuals have mild myopia and, as they have adequate near vision, they may not feel impaired.

Corneal pathology, the main preventable cause, was the second most common cause of blindness overall (11.8%), again similar to that found in the 1990 survey.⁴ There are many causes of corneal scarring, trachoma being one. Trachoma is still endemic in parts of the country, and a recent rapid assessment found that of 233 villages surveyed, 151 (64.8%) had individuals requiring trichiasis surgery.²² Pakistan has set up a dedicated national task force for trachoma control and is part of the GET 2020 alliance.²³

Phthisis/absent globe was the most important cause of unavoidable blindness and the third most important cause of unilateral SVI/BL. This highlights the importance of ocular trauma in Pakistan. A previous study on ocular trauma in NWFP found that 57.7% of cases had a perforating injury and that men outnumbered women by 5:1.²⁴ A similar gender difference in unilateral blindness was noted in our survey.

A comparison of the findings of this survey with others in the region shows that Pakistan has the lowest proportion of SVI/BL (57%) due to cataract (India 62.4%,²⁰ Bangladesh 82%¹²). Refractive error (including uncorrected aphakia) was lowest in Bangladesh (7.5%), followed by Pakistan (12%) and India (19.7%). Corneal opacity, responsible for nearly 1 in 10 cases of SVI/BL, was particularly prevalent in Pakistan but accounted for <1% of SVI/BL in India²⁰ and Bangladesh.¹² The reason for this high disparity is unclear but warrants further investigation. In subjects with <6/12, cataract and refractive error were of similar importance in Pakistan; however, this contrasts with Bangladesh where the main cause was cataract (73.4%), with refractive error only accounting for 18.9%.¹²

Our findings are markedly different from the findings in high income countries where the primary causes of blindness are age-related macular degeneration, diabetic retinopathy and myopic degeneration, one survey in the USA indicating that these causes accounted for 63% of blindness.²⁵

Based on the findings from this survey and future population dynamics, eye care service delivery needs to continue to expand in Pakistan, focusing principally on high-quality cataract surgery and aftercare, and on increased capacity for the correction of refractive errors. This recommendation is consistent with the prioritised areas of action for the region as outlined by the WHO South-East Asia policy VISION 2020: The Right to Sight. The more challenging conditions to control, namely glaucoma, macular degeneration and diabetic retinopathy, are also emerging as priorities.^26,27