Rapid Assessment of Avoidable Blindness in Kunming, China

Transcription

1 Rapid Assessment of Avoidable Blindness in Kunming, China Min Wu, MSc, 1 Jennifer L. Y. Yip, MBBS, MSc, 2,3 Hannah Kuper, ScD 2 Objective: To estimate the magnitude and causes of visual impairment (VI) in people aged 50 years in Kunming using the Rapid Assessment for Avoidable Blindness methodology, and to assess the prevalence of a pupillary defect in participants diagnosed as cataract visually impaired. Design: Population-based cross-sectional survey. Participants: We enrolled 2760 residents of Kunming prefecture in southwest China, 50 years of age. Methods: Forty-six clusters of 60 people were selected based on population proportional to size. Households from each cluster were selected using compact segment sampling (CSS) or quota sampling when CSS was not feasible. Visual acuity (VA) was assessed using a tumbling E chart. Lens status and cause of VI were determined by ophthalmologists using direct ophthalmoscopy through a dilated pupil where necessary. The pupillary reaction was assessed on undilated pupils when VI was detected. Main Outcome Measures: Prevalence of blindness (VA 3/60), severe VI (SVI) (VA 6/60), and VI (VA 6/18) using presenting VA (PVA). The causes of blindness and VI and prevalence of a pupillary defect in the cataract visually impaired were also assessed. Results: Of 2760 enumerated residents, 2588 were examined. The sample prevalence of bilateral blindness was 3.7% (95% confidence interval [CI], %). The prevalence of SVI was 3.0% (95% CI, %), and of VI was 9.1% (95% CI, %). The main cause of blindness was cataract (63.2% of blindness), followed by nontrachomatous corneal scar (14.7%), glaucoma (7.4%), and other posterior segment disease/neurologic disorders (4.2%). A pupillary defect was detected in 16% of those diagnosed with cataract VI. The cataract surgical coverage in the bilaterally blind was 58.9%, and 45% of operated eyes had good outcome with available correction (VA 6/18). The main barrier to cataract surgery was cost. Conclusions: Cataract remains the most important cause of preventable blindness in this poor region of China, and affordable provision of surgery would help to address this problem. Some cases of cataract blindness may not be preventable owing to preexisting comorbidity, as detected by the presence of a pupillary defect. Ophthalmology 2008;115: by the American Academy of Ophthalmology. Visual impairment (VI) is an important source of social burden worldwide. In 2002, global estimates suggest that there are 37 million people blind and 124 million with low vision, using World Health Organization (WHO) definitions with best-corrected visual acuity (VA). Over 90% of those visually impaired live in developing countries. A global initiative VISION 2020: The Right to Sight was launched in 1999 by the WHO and the International Agency for Prevention of Blindness with the aim of eliminating avoidable blindness by This program addresses disease control (prioritizing cataract, trachoma, childhood blindness/vitamin A deficiency, onchocerciasis, and refractive error/low vision), human resource development, and infrastructure development. The Originally received: March 27, Final revision: July 13, Accepted: August 1, Available online: October 22, Manuscript no Red Cross Hospital, Kunming, Yunnan, China. 2 International Centre for Eye Health, London School of Hygiene & Tropical Medicine, London, United Kingdom. 3 Department of Epidemiology, Institute of Ophthalmology, University College London, United Kingdom. targeted diseases account for 75% of total blindness and are treatable or preventable. The WHO estimated that the prevalence of blindness in people 50 years in China was 2.3% in A national survey conducted in China in 1987 indicated that the overall prevalence of blindness and low vision was 0.43% and 0.58%, respectively; in Yunnan Province, the prevalence of blindness and cataract was 0.6% and 0.66%, respectively, which was higher than the national average. 2 The main causes of blindness in Yunnan included cataract (41%), trachoma (18%), corneal opacities (19%), and glaucoma (6%). By 2020, an estimated 2.5 million people 60 years old in China will be blind from cataract. 3 Lack of up-to-date data on the prevalence of blindness is a major barrier to the Presented in part at: Association for Research in Vision and Ophthalmology Annual Meeting, May 2007, Fort Lauderdale, Florida. Supported by grants from Swire Foundation, London, United Kingdom; ORBIS International, London, United Kingdom; and Christian Blind Mission, Bensheim, Germany. Correspondence to Jennifer L. Y. Yip, MBBS, MSc, London School of Hygiene & Tropical Medicine, Keppel Street, London WC1E 7HT, United Kingdom. Jennifer.yip@lshtm.ac.uk by the American Academy of Ophthalmology ISSN /08/$ see front matter Published by Elsevier Inc. doi: /j.ophtha

2 Ophthalmology Volume 115, Number 6, June 2008 estimation of the global burden of blindness, particularly in China on account of its large population size. Yunnan is located in the southwest boundary of the People s Republic of China and is one of its poorest provinces. It borders Myanmar, Laos, and Vietnam, and has an area of 394,000 km 2, which accounts for 4.1% of the total area of China. The 2000 national census estimated that the population in Yunnan Province was and life expectancy was 68 years. Around 16.6% of the population was 50 years old and 76.6% were lived in a rural area. Kunming prefecture, the capital of Yunnan, has a similar demographic and urban/rural distribution as the whole of Yunnan. In 2000, Kunming had an estimated population of , of whom are 50 years old. The gross domestic product per capita in different administrative districts of Kunming ranged from $313 to $4324 and 28.7% of the population had gross domestic product per capita $625. Rapid assessment methods, such as Rapid Assessment of Cataract Surgical Services, are important tools for evaluation of ophthalmic care needs in different regions. This informs policy and allocation of limited resources. The Rapid Assessment of Cataract Surgical Services studies provide information on cataract blindness and services, and results are also used in extrapolations for global estimates of the burden of blindness. 1 The Rapid Assessment of Avoidable Blindness (RAAB) methodology updates Rapid Assessment of Cataract Surgical Services to include other avoidable causes of blindness, including trachoma, corneal blindness, onchocerciasis, and refractive error, 4 which form the main priority diseases of VISION However, this emphasis on anterior segment disease can detract attention from posterior segment diseases that may not be avoidable, but are nevertheless important causes of blindness. Moreover, using figures from these surveys to project global estimates of blindness can be misleading; patients with a dual diagnosis of VI will be preferentially diagnosed with avoidable conditions, with subsequent overestimation of the burden of disease treatable through cataract surgery and underestimation of diseases such as glaucoma, diabetic retinopathy, and age-related macular degeneration. This in turn could result in the misdirection of resources. A relative afferent pupillary defect (RAPD) is an established ophthalmic sign that represents asymmetrical disease of the optic nerve or retina, and is usually demonstrated in the eye with poorer VA. 5 A RAPD has also been reported in the contralateral eye with better vision of patients with asymmetrical cataracts. 6 However, a RAPD detected in patients with symmetrical cataracts indicates ocular comorbidity. In a blind patient with anterior segment disease, the presence of a RAPD suggests that the patient may also have posterior segment disease, which could be contributing to the poor VA detected. The aim of the study was to evaluate the magnitude and causes of VI in people ages 50 years in Kunming prefecture using the RAAB methodology, and to estimate the prevalence of a RAPD in participants diagnosed as cataract visually impaired. Materials and Methods The survey was carried out in 6 weeks between June and August 2006 in accordance with the Declaration of Helsinki. Ethical approval for the study was granted by the London School of Hygiene and Tropical Medicine and the Provincial Biological and Medical Ethical Committee of Kunming, China. Informed consent in the appropriate local dialect was obtained from all participants before examination. All participants diagnosed with cataract were referred to a local hospital for cataract surgery. The expected prevalence of blindness (WHO criteria: bestcorrected VA 3/60 in the better eye) in adults ages 50 years was estimated at 2.3%. 1 Given the confidence of 95% and a precision of 30% (the worst acceptable prevalence is 1.61%), a population size of adults ages 50 years in Kunming, a design effect (DEFF) of 1.4 for cluster size of 60, and 10% nonresponse, sample size of 2787 subjects was required. In total, 46 clusters of 60 adults ages 50 years were needed in this survey. A DEFF 1.4 for cluster size of 60 was selected based on experience from previous blindness surveys and to account for the similarities between individuals within each cluster. The population census data from 2000 were updated to 2006 data using the natural growth rate of the population, and the resulting population by enumeration area was used as the sampling frame. An enumeration area is an administrative area used by the census office and consists of a village or a suburb. The 50-yearold population was estimated for each enumeration area based on the census data. We used cluster sampling with a probability proportional to size of population to select 46 enumeration areas. First, a column showing the cumulative 50-year-old population was created in the sampling frame. The sampling interval was calculated as the total population size of 50-year-olds divided by the required number (46) of clusters. A random number between 0 and 1 was computer generated and the starting point for selection of the first cluster was determined by multiplying the random number with the sampling interval. The starting point was found in the cumulative column and the corresponding enumeration area, which contained the first number (starting point), was selected as the first cluster. The subsequent clusters were identified by adding the sampling interval to the previous number. This procedure was repeated until the last cluster (46th) was selected. Compact segment sampling (CSS) was chosen to select households within clusters in RAAB. 7 In CSS, selected enumeration areas were divided into segments, each with an approximately equal population and well-defined boundary. The number of segments equals the total 50-year-old population in the enumeration area divided by cluster size (60) so that each segment included about 60 people 50 years. All the segment numbers were written on small pieces of paper and one was randomly picked through lottery. Within the selected segment, the survey team visited each household door-to-door until 60 people 50 years were enumerated and those available and willing were examined. If there were households where there were known residents 50 years, but who were unavailable, this individual would be enumerated, and the survey team made at least two attempts to return and examine them. If all the households in a selected segment were visited and there were 60 enumerated eligible persons, a second segment was randomly selected to complete the cluster. However, it was not possible to visit each household by doorto-door in most urban areas and some rural areas of Kunming. Residents were worried about personal security and refused the survey team to approach the household even with an explanation from the village leader, the company of a village guide, and necessary documents, including an identity card of survey staff, introduction letter from the local hospital, and ethical approval. In 970

3 Wu et al Rapid Assessment of Avoidable Blindness some urban areas, there were high-rise building blocks with a secure front door and the team were refused entry to the building. Therefore, the survey teams used alternative sampling strategies to recruit eligible persons in 22 clusters where these circumstances prevailed. In a majority of these 22 clusters, eligible people were recruited from a community center. These centers were open to all, where elderly people from the administrative area could come to discuss administrative affairs or relax. In villages without a community center, the village leader would announce to 50-year-olds to come to a central location at a certain time for examination. To minimize potential bias in these 22 clusters, the survey team asked the village leader to inform all eligible local residents and not focus on people with eye diseases; in addition, the team also asked participants to encourage their spouses or eligible next of kin to participate in the survey. In these clusters, the survey team enumerated all people 50 years, and examined those who consented. This is a form of quota sampling, where participants in the cluster do not have a known probability of being selected but instead are selected based on the numbers or quota required. It should be emphasized that CSS is the accepted sampling strategy for RAAB, and should be employed in all future RAAB surveys. The alternative sampling strategy described here is not part of the RAAB methodology, and was used because of the specific circumstances encountered, which otherwise would have resulted in an unacceptably low participation rate. A standardized ophthalmic examination protocol for VA and identifying causes of VI was used for RAAB and has been described in detail elsewhere. 8,9 The main elements are described in brief. The definitions used in this study to calculate sample prevalences are blindness presenting VA (PVA) 3/60 in the better eye, severe VI (SVI) PVA 3/60 to 6/60 in the better eye, and VI PVA 6/60 to 6/18 in the better eye. The PVA is the VA determined based on the participants available correction. Visual Acuity Visual acuity for each eye was measured separately with a tumbling E chart using optotype size 6/18 on one side and size 6/60 on the other side at a 6- or 3-meter distance. The criteria for vision test at each level were 4 correct consecutive showings or 4 correct out of 5 consecutive showings. If the VA with available correction was 6/18 in either eye, then pinhole vision would also be measured following the same procedure. Ophthalmic Examination Using swinging flashlight method, RAPD was examined in participants with PVA 6/18 in either eye. Without dilation of the eyes, the participant was asked to stay in a dark place and look into the distance. The ophthalmologist used a bright torch to illuminate the retina of each eye separately from 15 below the visual axis. The torch was then swung from one eye to the other and illuminated each eye approximately 3 seconds, which was repeated 4 to 6 times. The reaction of the pupil during swinging was recorded. The lens status was assessed by torch or by distant direct ophthalmoscope by the ophthalmologist and the fundus and lens were examined using an ophthalmoscope in all participants diagnosed with VA 6/18. When pinhole VA of the examined eye was still 6/18, the pupil was dilated using short-acting 0.5% tropicamide eye drops (Santen Pharmaceutical Co. Ltd, Osaka, Japan), except eyes with a very shallow chamber, obvious severe cataract, large corneal opacity, or pupillary occlusion. All dilatated participants were offered 250 mg acetazolamide (ChenPai Pharmaceutical Co. Ltd., Jiangsu, China) because of the high risk of angle closure in this population. The barriers to cataract surgery and history of cataract surgery were also recorded. Two survey teams conducted the study. Training and validation of observers from both teams took place over 2 days, with a validation study where 20 of 46 participants had VI. There was good interobserver agreement for PVA, lens status, and main cause of VI, with (Kappa) values of 0.95, 0.96, and 0.98, respectively. A pilot study was conducted by the 2 teams before the start of data collection in one of the selected clusters to ensure all team members were familiar with the protocol and instruments. It was not possible to conduct a formal interobserver assessment of RAPD because few cases were identified with this condition. However, discussions were held on the standardization of methods to test for RAPD, and full agreement was reached on those patients who were evaluated in the validation study with RAPD. Statistical Analysis Data was analyzed using the RAAB package and STATA 9 (StataCorp, College Station, TX). Automated analyses in the RAAB package included unadjusted prevalence, principal cause, prevalence adjusted for DEFF, and age- and gender-adjusted prevalences. The associations between the prevalence of blindness, SVI, and VI and different predictors were analyzed using the chi-square test. Sensitivity analysis was performed to evaluate the difference in blindness estimates that may have resulted from the different sampling strategies. Results Overall, 2760 people were enumerated; 63 (2.3%) refused an examination and 109 (3.9%) were not available, leaving 2588 (93.8%) participants in the survey. Those who refused were more likely to be male (50.8% of those who refused vs. 40.3% of participants), but there was no difference in mean age between this group (65.8 years) and participants (65.5 years). People who were unavailable were younger (mean age, 63.6 years compared with 65.5 for participants). Of the 172 nonresponders, 5 were believed to be unilaterally blind and 3 bilaterally blind; 1 person was reported to have had cataract surgery. In the clusters where CSS was used, 1440 were enumerated; 1317 (91.5%) were examined, 85 (5.9%) were unavailable, and 38 (2.6%) refused. For the clusters with alternative sampling, 1320 were enumerated, 1271 (96.3%) were examined, 24 (1.8%) were unavailable, and 25 (1.9%) refused (Table 1). There were 95 participants who were blind, which provides an unadjusted sample prevalence of 3.7% (95% confidence interval [CI], 2.8% 4.6% with an observed DEFF of 1.6; Table 2). The prevalence of SVI was 3.0% (95% CI, 2.2% 3.8%; DEFF, 1.4), and of VI, 9.1% (95% CI, 7.5% 10.7%; DEFF, 2.1). All estimates of prevalence increased with age (Fig 1), but there were no difference in estimates between men and women. The age- and gender-adjusted prevalence of blindness (using PVA), SVI, and VI was 2.7%, 2.3%, and 7.2%, respectively. Extrapolating the ageand gender-adjusted sample prevalence estimates to the survey population area of Kunming district, there are an estimated people blind (95% CI, ) and (95% CI, ) with VI. People living in rural areas were more likely to be blind compared with urban dwellers (rural blindness 4.5% vs. 1.6% urban; odds ratio [OR], 2.9; 95% CI, ; P 0.01). There was also more blindness detected from CSS clusters compared to clusters with the alternative sampling strategy (4.3% vs. 3.0%, respectively; OR, 1.5; 95% CI, ; P 0.07). 971

4 Ophthalmology Volume 115, Number 6, June 2008 Table 1. Distribution of Sample Population by Age and Gender in Different Sampling Strategies Compact Segment Sampling Alternative Sampling Strategy Age Group Male, n (%) Female, n (%) Male, n (%) Female, n (%) (33.3) 315 (40.1) 111 (23.8) 271 (31.7) (30.7) 207 (26.3) 150 (32.2) 249 (29.2) (27.7) 194 (24.7) 154 (33.1) 259 (30.3) (8.3) 70 (8.9) 51 (11) 75 (8.8) Total Mean age (yrs) To evaluate how the different sampling methods affected our final prevalence estimates, we assessed the differences underlying the CSS and estimates from alternative sampling, and performed a sensitivity analysis to account for these. Participants selected using CSS sampling were younger (mean age for CSS, 64.7 years [95% CI, ] vs [95% CI, ] for alternative sampling; P 0.01), more likely to be male (45.4% males for CSS vs. 35.3% for alternative sampling; P 0.01), and mostly from rural areas (95.8% in CSS sample vs. 45.5% in alternative sampling; P 0.01). We used the age- and gender-adjusted blindness prevalence estimates from the alternative sampling group to extrapolate new figures using population structure from the CSS group. This reduced the estimated number of blind cases from 95 to 79, giving an overall blindness prevalence of 2.9%, which is within the range of the obtained estimates. The most common cause of blindness was untreated cataract, which was responsible for 63.2% of all blindness, followed by nontrachomatous corneal scar (14.7%). Glaucoma accounted for 7.4%, globe abnormality and phthisis for 6.3%, and other posterior segment disease/other neurologic disorders for 5.2% of blindness. Eighty-four percent of all blindness was considered to be due to avoidable causes (Table 3). Cataract was also the most common cause of SVI and VI (71.4% and 51.7%, respectively). Refractive error was the second most common cause of VI (36.0%), followed by other posterior segment disease/neurologic disorders (5.9%) and other corneal scar (4.7%). In this survey, 88.3% of SVI and 94.1% of VI was due to avoidable causes. The cataract surgical coverage (CSC) in bilaterally blind people was 58.9%, and in those with VA 6/60, 46.4%. There was no difference in CSC for each level of VA between men and women, and CSC for blind people was higher in urban areas (68.2%) compared with rural areas (53.2%). Of 91 people who had undergone cataract surgery, 32 received surgery in both eyes. Of the 123 operated eyes, intraocular lenses were implanted in 64.8%. After surgery, 38.8% of eyes had a poor outcome (VA 6/60 with available correction), which improved to 25.6% with best correction. The main causes of poor VA after cataract surgery were uncorrected refractive error (34.0%), ocular comorbidity (29.8%), and long-term sequelae of surgery (23.4%; Table 4). People visually impaired from cataract who had not undergone surgery cited that the most common reason for not having cataract surgery was that they could not afford it (36.4%), followed by being unaware of having a cataract (25.0%) and being told that they had a contraindication (13.6%). A pupillary defect was elicited in 66 of 408 (16%) participants with VA 6/18, of which the most common recorded principle cause of visual loss was cataract (37/66; 56%), followed by posterior segment disease. Fourteen of 66 people (21%) with a pupillary defect were blind. We also detected a pupillary defect in 10 out of 60 (16.6%) participants diagnosed with cataract blindness. Discussion Efficient and effective methods to estimate global blindness are vital to inform policy and resource allocation. The results reported here show that the prevalence of blindness (PVA 3/60 in the better eye) was 3.7% (95% CI, ). The adjusted prevalence of blindness was 2.7%, and using WHO definitions (best-corrected VA 3/60) was 2.3%, which is similar to the WHO estimates. 1 The estimates obtained here are higher than unadjusted prevalence estimates reported in previous surveys in other areas of China such as Shunyi county (2.8% in people 50 years; PVA 6/60), 10 Doumen county (2.7% in people 50 years; PVA 3/60), 11 Hong Kong (0.5% in people 60 years; PVA 6/60), 12 and Tibet (2.3% in people 50 years, PVA 6/60). 13 This may be because all previous survey areas except Tibet are in the more prosperous eastern areas of China compared with Kunming. Tibet currently has one of the highest cataract surgical rates in China, most likely owing to eye camps funded by foreign aid. A higher prevalence of blindness is associated with lower socioeconomic status 14 ; therefore, our findings are consistent with this notion. Yunnan is one of the poorest provinces in China; the results presented herein are likely to be more reflective of the western provinces of China. Rapid population Table 2. Unadjusted Prevalence of Blindness Estimates from Different Sampling Methods Compact Segment Sampling Alternative Sampling Strategy Overall Prevalence n % (95% CI) n % (95% CI) n % (95% CI) Blindness (PVA 3/60) ( ) ( ) ( ) SVI (PVA 6/60 and PVA 3/60) ( ) ( ) ( ) VI (PVA 6/18 and PVA 6/60) ( ) ( ) ( ) CI confidence interval; CSS compact segment sampling; NCSS noncompact segment sampling; PVA presenting visual acuity; SVI severe visual impairment; VI visual impairment. 972

5 Wu et al Rapid Assessment of Avoidable Blindness Table 4. Cataract Surgery Outcomes and Causes of Poor Outcome VA<6/60 VA<6/18 and VA>6/60 Causes n (%) n (%) Uncorrected refractive error 16 (34.0) 9 (45.0) Comorbidity 14 (29.8) 8 (40.0) Long-term sequelae 11 (23.4) 2 (10.0) Surgical complication 6 (12.8) 1 (5.0) All causes 47 (100) 20 (100) VA visual acuity. Figure 1. Prevalence of blindness, visual impairment (VI), and severe visual impairment (SVI) by age group. aging in China in recent years may also contributed to the higher estimates detected in our study. The higher prevalence estimates obtained in this survey could also be due to bias. We achieved good participation rates of 91% for CSS and 96% for the alternative sampling strategy, which suggests that the participants are likely to be representative of the survey population. However, logistical issues required the team to use 2 different types of sampling methods to select individuals within the cluster. Compact segment sampling is a well-recognized, objective method for household sampling, and it is likely to be less prone to bias than other methods used in rapid blindness surveys such as random walk. 15 This is the recognized sampling method for RAAB. In the alternative sampling strategy, people who were more healthy and mobile may have been more likely to participate because they were recruited from a center away from their homes. People who had some visual problems may also have been more likely to attend to have a free eye examination. Sensitivity analysis showed that accounting for differences in the structure of the population by age and gender suggested that the different sampling methods did not have a large effect on our final estimates. Any residual lower prevalence of blindness among the alternative sampling clusters compared with the Table 3. Causes of Blindness, Severe Visual Impairment (SVI), and Visual Impairment (VI) Blindness SVI VI Principal Causes n (%) n (%) n (%) Cataract 60 (63.2) 55 (71.4) 122 (51.7) Other corneal scar 14 (14.7) 5 (6.5) 11 (4.7) Glaucoma 7 (7.4) 0 (0) 0 (0) Globe abnormality and phthisis 6 (6.3) 1 (1.3) 0 (0) Other posterior diseases/cns 5 (5.2) 6 (7.8) 14 (5.9) disorders Uncorrected aphakia 2 (2.1) 3 (3.9) 2 (0.8) Surgical complication 1 (1.1) 2 (2.6) 0 (0) Refractive error 0 (0) 3 (3.9) 85 (36.0) AMD 0 (0) 2 (2.6) 2 (0.8) Avoidable causes 80 (84.2) 68 (88.3) 222 (94.1) Total number 95 (100) 77 (100) 236 (100) AMD age-related macular degeneration; CNS central nervous system. CSS clusters is likely to be attributable to the higher prevalence of blindness in rural than urban areas. Information bias is unlikely to have an effect on our estimates; the survey was carried out by experienced ophthalmologists with good interobserver kappa values. Extrapolated results show that there are an estimated people blind, with SVI, and with VI in Kunming prefecture. China is the most populous country in the world, and the predicted rapid aging of the population will increase the burden of age-related diseases. The CSC from this survey was lower than that reported in Tibet, 16 and there is a high proportion of patients with poor vision even after surgery. This suggests that more attention should be paid to improving the provision and quality of cataract surgery in this region. The presence of a RAPD in 16% of the cataract blind suggests that there is comorbidity that could be responsible for the poor vision. Operating on these cataracts may not produce good results unless a careful history and examination is performed and the appropriate eye is selected. There are reports of a paradoxical RAPD in patients with cataract in the eye with better vision, which resolves after cataract surgery. 6 In these cases, there is a large discrepancy in VA. In our survey, the cataract blind had vision of 3/60 in both eyes, the RAPD is unlikely to be due to the same mechanism, and most likely to represent optic nerve damage. Therefore using estimates from rapid surveys such as RAAB to project global figures for preventable blindness may be overly optimistic. Using results from Kunming, approximately would be cataract blind, of whom 3500 may not be preventable. Other causes of blindness that are irreversible and damage the optic nerve, such as glaucoma, may be responsible for a higher proportion of global blindness than present estimates suggest. Although RAPD is a useful tool to determine whether a patient or eye would benefit from cataract surgery, the training and standardization required may add complexities to the conduct of RAAB. A compromise between accuracy and efficiency occurs when resources are limited. More extensive surveys will yield more information at an extra cost. However, rapid surveys are efficient and effective in eliciting vital information, and remain an important tool for informing research and policy for blindness prevention. We have identified higher prevalence of blindness in Yunnan compared with other, more prosperous regions of China. The CSC estimated here, together with the high 973

6 Ophthalmology Volume 115, Number 6, June 2008 proportion of poor surgical outcomes, suggest that resources should be allocated to better surgical training and service provision in this region. References 1. Resnikoff S, Pascolini D, Etya ale D, et al. Global data on visual impairment in the year Bull World Health Org 2004;82: Zhang SY, Zou LH, Gao YQ, et al. National epidemiological survey of blindness and low vision in China. Chin Med J (Engl) 1992;105: Vision 2020 launched in the Western Pacific [press release]. Geneva: World Health Organization; September , Available at: html. Accessed May 22, Limburg H, Kumar R, Indrayan A, Sundaram KR. Rapid assessment of prevalence of cataract blindness at district level. Int J Epidemiol 1997;26: Thompson HS, Montague P, Cox TA, Corbett JJ. The relationship between visual acuity, pupillary defect, and visual field loss. Am J Ophthalmol 1982;93: Lam BL, Thompson HS. A unilateral cataract produces a relative afferent pupillary defect in the contralateral eye. Ophthalmology 1990;97: Turner AG, Magnani RJ, Shuaib M. A not quite as quick but much cleaner alternative to the Expanded Programme on Immunization (EPI) Cluster Survey design. Int J Epidemiol 1996;25: Wadud Z, Kuper H, Polack S, et al. Rapid assessment of avoidable blindness and needs assessment of cataract surgical services in Satkhira District, Bangladesh. Br J Ophthalmol 2006;90: Mathenge W, Kuper H, Limburg H, et al. Rapid assessment of avoidable blindness in Nakuru district, Kenya. Ophthalmology 2007;114: Zhao J, Jia L, Sui R, Ellwein LB. Prevalence of blindness and cataract surgery in Shunyi County, China. Am J Ophthalmol 1998;126: Li S, Xu J, He M, et al. A survey of blindness and cataract surgery in Doumen County, China. Ophthalmology 1999;106: Michon JJ, Lau J, Chan WS, Ellwein LB. Prevalence of visual impairment, blindness, and cataract surgery in the Hong Kong elderly. Br J Ophthalmol 2002;86: Dunzhu S, Wang FS, Courtright P, et al. Blindness and eye diseases in Tibet: findings from a randomised, population based survey. Br J Ophthalmol 2003;87: Dandona R, Dandona L. Socioeconomic status and blindness. Br J Ophthalmol 2001;85: Milligan P, Njie A, Bennett S. Comparison of two cluster sampling methods for health surveys in developing countries. Int J Epidemiol 2004;33: Bassett KL, Noertjojo K, Liu L, et al. Cataract surgical coverage and outcome in the Tibet Autonomous Region of China. Br J Ophthalmol 2005;89: