- Email: [email protected]
The Epidemiology of Trachoma in Rural Kenya
The Epidemiology of Trachoma in Rural Kenya Variation in Prevalence with Lifestyle and Environment Larry Schwab, MD, 1 Randolph Whitfield, Jr., MD, 2 Dennis Ross-Degnan, SeD, 3 Paul Steinkuller, MD, 4 Jack Swartwood, MPH, 5 the Study Survey Group* Purpose: Ocular examination surveys were carried out in Kenya by the International Eye Foundation as a component of the Kenya Rural Blindness Prevention Project to determine the national prevalence of blindness and ocular morbidity and major causes. A goal of the surveys was to determine the overall geographic distribution and severity of trachoma throughout Kenya. Methods: Using a random cluster household sampling technique, 13,803 people of all ages and of diverse cultural and ethnic backgrounds were identified in eight regions of Kenya. A detailed examination for active and inactive trachoma was carried out on each person surveyed as part of the general ocular examination. Results: The prevalence rate of visual loss (<20/60) due to trachoma in the better eye was 7.2/1000. Active trachoma was present in 19% of all persons examined, and 50% of all those with trachoma were found to have moderate to severe inflammation. Prevalence varied according to survey region from less than 1% in four regions where agriculture is the economic base, to 57% and 63% in two arid pastoral regions. Trachoma prevalence varied from 28% in children younger than 3 years of age to 11% in persons older than 60 years of age. Potentially blinding eyelid deformities secondary to chronic trachoma occurred in 5.0% of the rural population, and 1.2% of the rural population displayed associated corneal scarring. Lid scarring, corneal scarring, and lid deformities were greater in prevalence among females of all age groups when compared with males. Conclusions: Trachoma prevalence in Kenya varies widely from region to region. High prevalence is associated with high climatic aridity, and lower prevalence is asso ciated with areas of greater rainfall, sustainable agriculture, and a higher general standard of living. Within high-risk regions, there are wide variations in age-specific prevalence and severity of the disease. Potentially blinding sequelae of trachoma are more prevalent in females than in males. Ophthalmology 1995;102:475-482
Originally received: November 12, 1993. Revision accepted: October 12, 1994. 1
International Eye Foundation, Bethesda.
2
P. 0. Box 1323, Nyeri, Kenya. Department of Social Medicine, Harvard Medical School, Boston.
3 4
Department of Ophthalmology, Baylor College of Medicine, Houston.
5
Citizens Network for Foreign Affairs, Washington, DC.
* Members of the Study Survey Group are listed in the Appendix at the end of this article.
Study was completed with the cooperation and financial assistance of the Government of Kenya, Nairobi, Kenya, the United States Agency for International Development, Washington, DC, the World Health Or ganization Programme for the Prevention of Blindness, Geneva, Swit zerland, the Kenya Society for the Blind, Nairobi, Kenya, and institutional and private donors to the International Eye Foundation, Bethesda, Maryland. The authors have no proprietary interest in the surveys, nor in the equip ment or diagnostic medicaments used in the work. Reprint requests to Larry Schwab, MD, International Eye Foundation, 7801 Norfolk Ave, Bethesda, MD 20814.
475
Ophthalmology
Volume 102, Number 3, March 1-995
Trachoma is the major preventable cause of blindness in areas of sub-Saharan Africa where onchocerciasis is not endemic. Although there have been several trachoma prevalence studies in the region published in the last de cade,l-7 there have been few studies of the disease as a national problem, either in terms of its importance as an overall cause of blindness8 or its variation in prevalence and severity within a country. This study describes the epidemiology of trachoma in Kenya as determined by eye disease surveys conducted between 1977 and 1981 by the International Eye Foun dation as a component of the Kenya Rural Blindness Pre vention Project. 9 The overall goal of this project was to strengthen the curative and preventive services ofthe rural eye care system. The surveys were to investigate differences in the prevalence and causes of visual loss and eye disease in rural Kenya, which enabled more rational planning of eye care services through the Kenya Ministry of Health. Situated astride the equator on the East African coast, Kenya is a land ofgreat geographic and cultural diversity. Its habitats include fertile and temperate highlands, hot and humid coastal plains, semi-arid grassland plateaus, and great expanses ofarid bush. People inhabit every area of the country, but the current population of more than 30 million is concentrated primarily in the 15% of the total land area which is agriculturally productive. Kenya's population consists of more than 30 major and 40 minor tribes, all with different histories, languages, and traditions. Unlike the cities where tribal groups have mixed, most areas in rural Kenya maintain a high degree of ethnic and cultural homogeneity. Traditional lifestyles vary widely according to socioeconomic determinants, local agricul ture potential, and environment. There is wide intra country variation in level of economic development and access to health services. Trachoma is an infection that is influenced greatly by culture, geography, and level of economic develop ment.2·7·10·11 Because it isacommunicabledisease ofrather low infectivity, it requires high frequency and duration of exposure to persist in a population. According to the World Health Organization, trachoma remains a major cause of preventable blindness. Blindness results with chronic and repeated infection and corneal scarring resulting from en tropion and trichiasis. Entropion, trichiasis, secondary cor neal ulceration, pyogenic infection, and blindness are the result of lifelong trachoma (Fig I). Because transmission is affected by community access to water supply/· 11 by environment and personal hy giene/·12 or by sociocultural customs such as cattle herd ing, 7 which often are associated with economic develop ment, the prevalence and nature of the disease may un dergo variation and decline with national economic development. Trachoma is known to be a major cause ofocular mor bidity and visual loss in Kenya. In previous analyses from the current study, we reported that the ocular sequelae of trachoma were responsible for 19% of all blindness (visual acuity less than 20/400 in the better eye) and 13% of visual impairment (visual acuity less than 6/18 but better than or equal to 20/400) in · Kenya. 13 Loewenthal and Pe'er3
476
Figure 1. An adult woman with entropion, trichiasis, corneal ulceration, pyogenic infection, and blindness due to lifelong trachoma.
found the prevalence of inflammatory trachoma to be 42.8% in the remote northern Turkana region of Kenya, whereas Whitfield et al 14 reported a prevalence of 80.5% among the pastoral Samburu people.
Methods Sample Design Ophthalmic clinical examination surveys were conducted in eight rural districts of Kenya. A probability sample of the national rural population was not included in the de sign protocol because of constraints in time and resources. Most of Kenya's land surface is only lightly populated, and rural regions where population density was greatest were chosen for study. These rural areas, which repre sented major population groups and the ranges ofecologic conditions that might impact on eye disease, were chosen to be surveyed. Using simplifying assumptions about proportions of population represented in each survey, national estimates were imputed from independent regional samples. It was assumed that lifestyles of the majority of people in each
Schwab et al · Trachoma in Rural Kenya region were similar from household to household and vil lage to village, an assumption considered valid because of the homogeneity of tribal life in Kenya. In addition, it was assumed that Kenya census data were valid to a rea sonable degree, considering the enormous constraints in gathering census data in an African country where 80% of its people live in rural areas. Characteristics of regional surveys are presented in Table 1. The political boundaries of administrative districts surveyed are delineated in the Government of Kenya Atlas ( 1978) and have not been altered since the publication of the atlas. The methodology used to draw the samples for these surveys has been described elsewhere. 13 · 15 To summarize the methodology, a random cluster sample similar to the one currently recommended by the World Health Orga nization for immunization surveys 16 was drawn from se lected administrative divisions using a sampling frame based on tax lists, land adjudication records and, in less settled areas, lists of heads of household assembled by local chiefs. Although sample size was sometimes varied to meet par ticular objectives, each survey generally was planned to con sist of 30 clusters of households. Each cluster was begun at a starting point selected at random from the sampling frame and was continued by adding all members ofthe next nearest household until the total number ofeligible residents in the cluster exceeded 60 individuals. Examinations in each cluster
continued until at least 90% of all eligible persons were ex amined. Thus, there were approximately 1800 individuals in the typical district survey, and a total of 13,803 persons were examined in eight surveys. Precise population distribution data are not available for Kenya, but the Government of Kenya Atlas (1978) states that 90% of Kenya's population lives on only 15% of the country's land surface, mainly in the southwestern and central Kenya highlands where land is arable and agriculturally productive. Only 10% of Kenya's population lives on 85% of the land in subsistence and pastoral life styles. This vast area of Kenya is generally nonarable. The plain, unhatched area on the Kenya map (Fig 2) represents these arid lands. Because of the enormous logistical dif ficulties in surveying peoples in those areas of low-pop ulation density, the survey was designed to study popu lations living in similar circumstances but that were more accessible logistically. Results from these areas then were used to estimate trachomatous rates for those peoples liv ing in the low-population density areas that were not sur veyed. Areas selected for the survey were: Baringo, Ka jiado, Kakamega, Kisii, Kwale, Meru, Nyanza, and Nyeri.
Clinical Methods In the first five site surveys, examinations were conducted at local facilities (schools, churches, health facilities) con-
Table 1. Characteristics of Regional Survey Areas Sample Size
Estimate% of Rural Population
Meru(Meru)
1142
16.1
Hills
Nyeri (Kikuyu)
1825
20.4
Mt. Kenya highlands
Kwale (Mijikenda)
1342
6.8
Kakamega (Luhya)
1651
23.6
Plateau
Nyanza (Luo)
1807
11.5
Lake Victoria basin
Kisii (Kisii)
1753
6.2
District (tribal group)
Baringo (Tugen)
702
Geography (altitude)
Coastal plain
Southwestern plateau Tugen hills
6.7
(Tugen)
480
(Njemps)
581
Coastal plain
(Pokot)
596
Edge of desert
Kajiado (Maasai)
1924
8.7
Plateau
Grassland plateau
Climate (rainfall)* Temperate to hot (moderate) Cool, temperate (high) Hot, humid, (moderate) Temperate to hot (moderate) Hot, humid (moderate) Temperate (high) Temperate to hot (moderate) Hot (moderate to low) Hot (arid, but near irrigation) Very hot (arid, semi-desert) Hot (arid, semi desert)
Economy* Coffee, tea, "chat" cultivation; subsistence farming Coffee and tea farming; subsistence farming Fishing; tourism; subsistence farming Cotton farming; some coffee farming; subsistence farming Cotton farming; fishing; subsistence farming Tea and coffee farming; subsistence farming Coffee farming; subsistence farming Subsistence farming; herding livestock Herding livestock; some subsistence farming Herding livestock Herding livestock
• Specific rainfall data and regional economic indicators data are not available from published sources or from the government of Kenya.
477
Ophthalmology
Volume 102, Number 3, March 1995
Survey Prevalence of Trachoma Area 62.6% 1. Meru 10.7% 2. Nyeri 0.8% 3. Kwale 0.1% 4. Kakamega 0.2% 5. Nyanza 0.0% 6. Kisii 29.2% 7. Baringo 56.8% 8. Kajiado
Figure 2. Survey areas, pop ulation density, and preva lence of active trachoma.
Population (per sq. km.)
D
venient to the residents of the clusters. Fewer randomly selected survey members were lost to the study when in dividuals were examined at home; survey members were sometimes reluctant to travel any distance away from their households to be examined. To simplify logistics and im prove coverage, survey members were examined in their own households in the remaining three surveys. Each sur vey team consisted ofa minim urn ofan ophthalmologist, an ophthalmic clinical officer or ophthalmic assistant, a registrar, and a locally hired clerk who was familiar with the survey area and local language. Each survey subject was registered, and visual acuity was measured in each eye. An ophthalmic history was taken by the clinical of ficer. All survey members then were examined by the ophthalmologist. Ophthalmologists were members of either the staff of the Kenya Rural Blindness Prevention Project or the International Eye Foundation, or they worked for the Kenya Ministry of Health. There were nine separate examiners over the course of the eight surveys, but the three resident Kenya Rural Blindness Prevention Proj ect ophthalmologists examined 69% of the total number of study subjects. Each examination was carried out according to a structured clinical protocol, with a focal handlight, X4 loupes, and an ophthalmoscope. The examination consisted of a corneal evaluation and examination of the bulbar and tarsal conjunctiva. Every survey subject was examined for evidence of current trachomatous infection or for evidence of previous infection (scars on the tarsal conjunctiva, pannus, Herbert pits). Tra choma was scored according to recommendations of the World Health Organization 17 current at the time of the surveys using the following indicators and cri teria (Table 2).
478
=under20
[[ill
=21 to 149
~
= 150 and over
Methods of Analysis Because of the great sensitivity of estimates of the prev alence and causes of visual loss to the distribution of age in a population, a direct standardization procedure was carried out to minimize the effects of random differences in the age distribution in each of the eight regional survey samples. Each sample population was adjusted to the na tional population age distribution from the 1979 census by weighing age categories according to whether they were over- or under-represented relative to the standard. This procedure allows more appropriate regional comparisons and ensures that national estimates are less influenced by sampling differences. Because the survey districts and the locations within them were not selected randomly, there is no unbiased procedure for constructing true population estimates of ocular conditions. Within each region, it was assumed that the particular locations from which the sample was randomly drawn were representative of a broader popu lation group and that estimates of rates from the survey sample would apply to the underlying regional population within reasonable bounds of error. National estimates were computed by attributing to each regional survey a certain proportion of the national rural population based on similarity in tribal composition, geography, and climate (Table 1). Observations within each survey were appropriately weighted so that the na tional estimates reported reflect these attributed propor tions. For these reasons, and because the sample was drawn in clusters, there is no unbiased method to calculate sam pling errors or confidence intervals for the prevalence es timates. No standard errors are reported, although it would be possible to compute rough estimates based on as
Schwab et al · Trachoma in Rural Kenya Table 2. Criteria for Grading Trachoma in Rural Kenya Survey Criteria
Indicator Intensity of Inflammation Mild Moderate Severe Inactive
At least 5 follicles in superior tarsal plate zones 1 and 2, but fewer than 5 mature follicles 5 or more mature follicles in zones 1 and 2 with or without papillary hypertrophy Marked papillary hypertrophy with or without follicles Lid scarring, Herbert's pits, pannus without any signs of current inflammation
Type and Severity of lid Scarring Mild Moderate Entropion Trichiasis
Fine linear scars in zone 2 Sheet scarring in zone 2 with no lid deformity Deformity of lid with no lashes abrading conjunctiva or cornea Lid deformity and lashes abrading conjunctiva or cornea
Type and Severity of Corneal Lesion Leukoma 1 Leukoma 2 Leukoma 3 Keratitis
Corneal scar with no effect on visual acuity Scar causing slight vision loss Scar across the visual axis Punctate corneal lesions due to trachoma
sumptions of simple random sampling from the infor mation on sample size presented and to inflate them by an appropriate factor (perhaps 2-4 times 18) to account for the nonrandomized design and clustering.
Results An estimated 19% of rural Kenyans have active trachoma, and 12% not currently infected have sequelae of previous infections. Of persons who have clinical signs ofexposure to current or previous trachoma, approximately one in six have deformed eyelids as a result, placing them at high risk of trachoma-induced visual loss, whereas I in 44 have already had some visual loss due directly to the disease. The occurrence of trachoma in rural Kenya is by no means a uniform phenomenon (Fig 3). Overall prevalence oftrachoma inflammation was 63% and 57% in the two most severely affected survey regions, 29% and 11% in
Table 3. Prevalence of Trachoma by Survey Area*
Prevalence
100%.----------------------------------------, 80% 60% 40% 20% 0%
0-4
two others, and less than 1% in the remaining four. Tra choma was most likely to be endemic in arid areas oflow population density, including Baringo, Kajiado, and Meru. Trachoma was less prevalent in areas of denser settlement. Both cash crop and subsistence agriculture is practiced in those areas ofdenser settlement where rainfall is higher and sufficient to sustain it. It also was found to be present but less prevalent in areas of dense human settlement or with higher annual rainfall. Table 3 displays the wide variation of prevalence rates by regional survey site. Figure 3 presents the distribution of trachoma grades in these areas by age cohort in regions where prevalence of active disease is greater than 1%. Overall rates of in flammation peak in the first decade (in fact, in the first 3 years) and then decline rapidly to approximately 20 years ofage and continue to decline more gradually throughout adulthood to old age. In each age group, approximately 50% ofall trachomatous inflammation is classified as mild,
10-14 20-24 30-34 40-44 50-54 60-64 70-74 5-9 15-19 25-29 35-39 45-49 55-59 65-69 75+
Age Group Figure 3. Trachoma prevalence by age group in rural Kenya. Solid bars = mildly active trachoma; shaded bars = moderately to severely active trachoma; open bars = signs of past trachoma.
Survey Area Meru Nyeri Kwale Kakamega Nyanza Kisii Baringo Kajiado
Prevalence of Trachoma (%)
62.6 10.7 0.8 0.1 0.2 0.0 29.2 56.8
* These data correlate by number with survey sites displayed in Fig ure L
479
Ophthalmology
Prevalence
Volume 102, Number 3, March 1995
FEMALES
MALES Prevalence
100%.-----------------------.
Figure 4. Sequelae of tra choma by sex and age group in rural Kenya. Solid bars = corneal scarring; shaded bars = potentially blinding lid de formity; open bars = lid scar ring of any severity.
80%
80%
60%
60%
40%
40%
20%
20%
0%
.__~--
0-9 20-29 40-49 60.09 10-19 30-39 50-59 70 +
0-9 20-29 40-49 60.09 10-19 30-39 50-59 70 +
Age Group
Age Group
and 50% is moderate to severe. The age-specific prevalence among those with any indication of either present or pre vious trachoma reaches 80% to 90% by the fifth decade. Slightly to moderately higher prevalence of infection is seen in females than in males in every group, with the relative risk varying from 1.0 to 2.0. This increased ex posure in females is responsible for higher rates of tra chomatous lid scarring in women, especially of dense scarring causing lid deformities (Fig 4). In addition, women show higher rates of related visually disabling le sions such as corneal leukoma and punctate keratitis. Within the geographically circumscribed and relatively homogeneous areas of each survey, the prevalence of in flammatory trachoma and trachomatous lid scarring is widely variable. This is demonstrated by the within-cluster rates in each region (Fig 5). Although variations are wide, the within-cluster rates tend, nonetheless, to fall within characteristic ranges in each survey area, except in Bar ingo, which was the only survey planned to take into ac count wide ecologic and cultural variability. Rates ofcurrent infection within sample clusters tended to vary with presence of trachoma lid scarring (r = 0.88, P < 0.05), which indicates a continuity between historic and current factors relating to trachoma. The relation is not completely consistent, and in two areas (Nyeri and Nyanza) the average rate oflid scarring is relatively greater than might be expected, given the current prevalence of active disease. Severe lid deformities (entropion with tri chiasis), which greatly increase the risk of visual loss, are the result of repeated re-infection. The cluster-specific prevalence of these severe deformities in all endemic areas increases significantly in proportion to the rate of active inflammation (r = 0.79, P < 0.05). Just as overall prevalence rates vary within and across survey regions, the age-related profile of the disease also varies (Fig 6). In Kajiado, Meru, and Nyeri, the peak prevalence of inflammatory trachoma occurs in children younger than 5 years of age with consistently lower rates
480
100% . - - - - - - - - - - - - - - . ,
in older age groups. In Baringo, there are local prevalence peaks in children younger than 5 years of age and lower rates in older children. Rates of trachomatous infection in Baringo tend to be higher in adults, and those individ uals who are older than 40 years of age are the most se verely affected. These findings parallel those in a recent village trachoma study in Tanzania. 19 The age-associated pattern of trachomatous infection varies within areas of relatively uniform cultural practices and environmental conditions. In three sampling locations in Kajiado, children younger than 5 years of age had high rates of infection, and rates were relatively constant at much lower levels in older age groups. In the other three sampling areas, rates of infection were very high among children younger than 5 years of age, remained at much higher levels among older children and adolescents, and declined slowly with age. Prevalence 100%~------------------------------------.
Meru
Kwale (23)
(30)
Nyeri (30)
Nyanza (30)
Kakamega (29)
Baringo (39) Kisii
(30)
Kajiado (34)
Survey Area (Number of Clusters)
Figure 5. Variation in prevalence of active trachoma in 60-person sample clusters.
Schwab et al · Trachoma in Rural Kenya
NYERI
Prevalence
MERU
Prevalence
10~kr---------------------------~
60% 40% 20%'-····.,.~·-·········
Oo/oL_~~~==~~~~~==:::=:J 0-2
3-5
S.B
12-14 18-20 31-40 51-60 9-11 15-17 21-30 41-50 61+
S.B
0%0-2 3-5
Age Group
Age Group
BARINGO
Prevalence
KAJIADO
100% ,----------------------------~ 80%
12-14 18-20 31-40 51-60 9-11 15-17 21-30 41-50 61+
......................-··
Figure 6. Prevalence of active trachoma in endemic survey areas by age group.
Prevalence
100% r---------------------------..,
·.............
-1----------------------,~~---1
60o/o f-->·o::···-=··:-.:------:=---4----------,;::,-----l ········~······· ~
.......
--
... ,.
40%-1---------------------~--~---1
./ 20o/o t"-o;;;;;:;:'~~.,=-=---::/,._-~,.,-------------j 0%
-·-·-·---~-·-·-·-·-·-·-·-··
0-2
Pokot
3-5
S.B
12-14 18-20 31-40 51-60 9-11 15-17 21-30 41-50 61+
Age Group Plains Tugen Njemps Hill Tugen
Discussion Trachoma remains a major cause of preventable blindness in Africa and in the world. Its endemic status and severity are in part determined by level of economic development, access to clean water, availability of adequate sanitation, and access to preventive and curative health services. Be cause trachoma is a communicable disease of relatively low infectivity, it requires high frequency and duration of exposure to persist in a population. Transmission of the disease at the community level can be interrupted by improved personal hygiene and by provision of an ade quate supply of clean water. Prevalence and severity of infection can decrease markedly as a community under goes socioeconomic development. 20 The results of the Kenya blindness surveys support these findings. Kenyans who live in the arid and semi-arid areas of the country primarily are involved in pastoral activity. These groups are represented in the surveys by the Maasai and Kalenjin nomads and by the semi-nomads ofBaringo district. Most of the people in these areas live outside Kenya's money economy at a lower standard of living than the more settled population groups. It is among these nomadic groups that the prevalence of active trachoma is the highest. An exception is Meru district, where trachoma prev alence was the highest of any district surveyed, despite the fact that rainfall is moderate and considerable cash is generated by the cultivation of tea, coffee, and "chat," a leaf which is chewed as a stimulant, for which there is a
12-14 18-20 31-40 51-60 9-11 15-17 21-30 41-50 61+
Age Group
regional cash market in Kenya and neighboring Somalia. Despite its moderate rainfall, the pumaceous basaltic soil in Meru permits such rapid runoff of rain from this high land area that water remains relatively inaccessible. Al though the people of Meru are not nomadic, the district itself is nevertheless remote and underserved. There are fewer community development projects that could elevate the standard ofliving in Meru District than in other survey districts with similar settled populations. The Meru area is more similar to the arid districts in both ecologic profile and trachoma prevalence because of these factors. It is instructive to contrast the relative prevalence rates of active versus cicatricial trachoma among districts. Lid scarring from inactive trachoma was found to be more prevalent than active disease in the Nyanza and Nyeri districts and to a lesser degree in the Kisii district. In these areas, commercial agriculture has been developed to a greater extent due to favorable rainfall, and, as a conse quence, the standard of living is higher than in districts where active trachoma was found to be more prevalent. The presence of healed scars indicates that trachoma was prevalent within the lifetimes of older people, but eco nomic development and access to improved health ser vices have caused the disease to disappear as a serious public health problem.
Summary Inflammatory trachoma remains a serious problem mainly in areas of rural Kenya where arid climatic con
481
Ophthalmology
Volume 102, Number 3, March 1995
ditions and poor access to water exist and where nomadic lifestyles prevail. In some areas where inflammatory tra choma does not exist, there is evidence of historic disease in older age cohorts. In those areas, economic develop ment, improvements in sanitation, greater access to clean water, and access to health services appear to have con tributed to the elimination of trachoma. From the public health planning perspective, emphasis on eye health ser vices in those areas must be on prevention of blindness due to the sequelae of trachoma rather than elimination of infection.
Appendix Members of the Study Survey Group H. Adala, N. Bakker, H. Beekhuis, G. Bisley, N. Dekkers,
I. Hoshiwara, H. Mandalia, P. Mandalia, R. Meaders,
J. 0. Ochola, and V. Sheffield.
References 1. Sukwa TY, Ngalande TC, Mwandu DH, et al. Prevalence and distribution oftrachoma in the Luapula Valley, Zambia. East Afr Med J 1992;69:34-6. 2. Katz J, Zeger SL, Tielsch JM. Village and household clus tering of xerophthalmia and trachoma. Int J Epidemiol 1988;17:865-9. 3. Loewenthal R, Pe'er J. A prevalence survey of ophthalmic diseases among the Turkana tribe in north-west Kenya. Br J Ophthalmol 1990;74:84-8. 4. De Sole G. Impact of cattle on the prevalence and severity of trachoma. Br J Ophthalmol 1987;71 :873-6. 5. Kabo AM, Di M. La trachome dans le Departement de Zinder (Niger). Rev Int Trach Pathol Ocul Trop Subtrop Sante Publique 1989;66:43-54. 6. Huguet P. Le trachome au Niger: (resultats d'une enquete sondage de Tahoua). Rev Int Trach Pathol Ocul Trop Sub trop Sante Publique 1989;66:31-41.
482
7. Tielsch JM, West KP Jr, Katz J, et al. The epidemiology of trachoma in southern Malawi. Am J Trop Med Hyg 1988;38:393-9. 8. Faal H, Minassian D, Sowa S, Foster A. National survey of blindness and low vision in The Gambia: results. Br J Ophthalmol1989;73:82-7. 9. Arbuckle RD. The International Eye Foundation/Kenya Rural Blindness Prevention Project. Soc Sci Med 1983;17: 1789-92. 10. Bailey R, Osmond C, Mabey DC, et al. Analysis ofhousehold distribution oftrachoma in a Gambian village using a Monte Carlo simulation procedure. Int J Epidemiol 1989; 18:944 51. II. Prost A, Negrel AD. Water, trachoma and conjunctivitis. Bull World Health Organ 1989;67:9-18. 12. Taylor HR, West SK, Mmbaga BBO, et al. Hygiene factors and increased risk of trachoma in central Tanzania. Arch Ophthalmol 1989;107:1821-5. 13. Whitfield R, Schwab L, Ross-Degnan D, et al. Blindness and eye disease in Kenya: ocular status survey results from the Kenya Rural Blindness Prevention Project. Br J Ophthalmol 1990;74:333-40. 14. Whitfield R Jr, Schwab L, Bakker NJA, et al. Cataract and corneal opacity are the main causes ofblindness in the Sam bum tribe of Kenya. Ophthalmic Surg 1983; 14:139-44. 15. Ross-Degnan D, Schwab L, Mburu FM. Field methodology for ocular surveys in rural Africa. Soc Sci Med 1983;17: 1793-6. 16. Henderson RH, Sundaresan T. Cluster sampling to assess immunization coverage: a review of experience with a sim plified sampling method. Bull World Health Organ 1982;60: 253-60. 17. Dawson CR, Jones BR, Tarizzo ML. Guide to Trachoma Control in Programmes for the Prevention of Blindness. Geneva: World Health Organization, 1981. 18. Tabbara KF, Ross-Degnan D. Blindness in Saudi Arabia. JAMA 1986;255:3378-84. 19. Turner VM, West SK, Munoz B, et al. Risk factors for tri chiasis in women in Kongwa, Tanzania: a case-control study. Int J Epidemiol 1993;22:341-7. 20. Meredith SJ, Peach HG, Devanesen D. Trachoma in the Northern Territory of Australia, 1940-1986. Med J Aust 1989;151:190, 192, 194-9.
Originally received: November 12, 1993. Revision accepted: October 12, 1994. 1
International Eye Foundation, Bethesda.
2
P. 0. Box 1323, Nyeri, Kenya. Department of Social Medicine, Harvard Medical School, Boston.
3 4
Department of Ophthalmology, Baylor College of Medicine, Houston.
5
Citizens Network for Foreign Affairs, Washington, DC.
* Members of the Study Survey Group are listed in the Appendix at the end of this article.
Study was completed with the cooperation and financial assistance of the Government of Kenya, Nairobi, Kenya, the United States Agency for International Development, Washington, DC, the World Health Or ganization Programme for the Prevention of Blindness, Geneva, Swit zerland, the Kenya Society for the Blind, Nairobi, Kenya, and institutional and private donors to the International Eye Foundation, Bethesda, Maryland. The authors have no proprietary interest in the surveys, nor in the equip ment or diagnostic medicaments used in the work. Reprint requests to Larry Schwab, MD, International Eye Foundation, 7801 Norfolk Ave, Bethesda, MD 20814.
475
Ophthalmology
Volume 102, Number 3, March 1-995
Trachoma is the major preventable cause of blindness in areas of sub-Saharan Africa where onchocerciasis is not endemic. Although there have been several trachoma prevalence studies in the region published in the last de cade,l-7 there have been few studies of the disease as a national problem, either in terms of its importance as an overall cause of blindness8 or its variation in prevalence and severity within a country. This study describes the epidemiology of trachoma in Kenya as determined by eye disease surveys conducted between 1977 and 1981 by the International Eye Foun dation as a component of the Kenya Rural Blindness Pre vention Project. 9 The overall goal of this project was to strengthen the curative and preventive services ofthe rural eye care system. The surveys were to investigate differences in the prevalence and causes of visual loss and eye disease in rural Kenya, which enabled more rational planning of eye care services through the Kenya Ministry of Health. Situated astride the equator on the East African coast, Kenya is a land ofgreat geographic and cultural diversity. Its habitats include fertile and temperate highlands, hot and humid coastal plains, semi-arid grassland plateaus, and great expanses ofarid bush. People inhabit every area of the country, but the current population of more than 30 million is concentrated primarily in the 15% of the total land area which is agriculturally productive. Kenya's population consists of more than 30 major and 40 minor tribes, all with different histories, languages, and traditions. Unlike the cities where tribal groups have mixed, most areas in rural Kenya maintain a high degree of ethnic and cultural homogeneity. Traditional lifestyles vary widely according to socioeconomic determinants, local agricul ture potential, and environment. There is wide intra country variation in level of economic development and access to health services. Trachoma is an infection that is influenced greatly by culture, geography, and level of economic develop ment.2·7·10·11 Because it isacommunicabledisease ofrather low infectivity, it requires high frequency and duration of exposure to persist in a population. According to the World Health Organization, trachoma remains a major cause of preventable blindness. Blindness results with chronic and repeated infection and corneal scarring resulting from en tropion and trichiasis. Entropion, trichiasis, secondary cor neal ulceration, pyogenic infection, and blindness are the result of lifelong trachoma (Fig I). Because transmission is affected by community access to water supply/· 11 by environment and personal hy giene/·12 or by sociocultural customs such as cattle herd ing, 7 which often are associated with economic develop ment, the prevalence and nature of the disease may un dergo variation and decline with national economic development. Trachoma is known to be a major cause ofocular mor bidity and visual loss in Kenya. In previous analyses from the current study, we reported that the ocular sequelae of trachoma were responsible for 19% of all blindness (visual acuity less than 20/400 in the better eye) and 13% of visual impairment (visual acuity less than 6/18 but better than or equal to 20/400) in · Kenya. 13 Loewenthal and Pe'er3
476
Figure 1. An adult woman with entropion, trichiasis, corneal ulceration, pyogenic infection, and blindness due to lifelong trachoma.
found the prevalence of inflammatory trachoma to be 42.8% in the remote northern Turkana region of Kenya, whereas Whitfield et al 14 reported a prevalence of 80.5% among the pastoral Samburu people.
Methods Sample Design Ophthalmic clinical examination surveys were conducted in eight rural districts of Kenya. A probability sample of the national rural population was not included in the de sign protocol because of constraints in time and resources. Most of Kenya's land surface is only lightly populated, and rural regions where population density was greatest were chosen for study. These rural areas, which repre sented major population groups and the ranges ofecologic conditions that might impact on eye disease, were chosen to be surveyed. Using simplifying assumptions about proportions of population represented in each survey, national estimates were imputed from independent regional samples. It was assumed that lifestyles of the majority of people in each
Schwab et al · Trachoma in Rural Kenya region were similar from household to household and vil lage to village, an assumption considered valid because of the homogeneity of tribal life in Kenya. In addition, it was assumed that Kenya census data were valid to a rea sonable degree, considering the enormous constraints in gathering census data in an African country where 80% of its people live in rural areas. Characteristics of regional surveys are presented in Table 1. The political boundaries of administrative districts surveyed are delineated in the Government of Kenya Atlas ( 1978) and have not been altered since the publication of the atlas. The methodology used to draw the samples for these surveys has been described elsewhere. 13 · 15 To summarize the methodology, a random cluster sample similar to the one currently recommended by the World Health Orga nization for immunization surveys 16 was drawn from se lected administrative divisions using a sampling frame based on tax lists, land adjudication records and, in less settled areas, lists of heads of household assembled by local chiefs. Although sample size was sometimes varied to meet par ticular objectives, each survey generally was planned to con sist of 30 clusters of households. Each cluster was begun at a starting point selected at random from the sampling frame and was continued by adding all members ofthe next nearest household until the total number ofeligible residents in the cluster exceeded 60 individuals. Examinations in each cluster
continued until at least 90% of all eligible persons were ex amined. Thus, there were approximately 1800 individuals in the typical district survey, and a total of 13,803 persons were examined in eight surveys. Precise population distribution data are not available for Kenya, but the Government of Kenya Atlas (1978) states that 90% of Kenya's population lives on only 15% of the country's land surface, mainly in the southwestern and central Kenya highlands where land is arable and agriculturally productive. Only 10% of Kenya's population lives on 85% of the land in subsistence and pastoral life styles. This vast area of Kenya is generally nonarable. The plain, unhatched area on the Kenya map (Fig 2) represents these arid lands. Because of the enormous logistical dif ficulties in surveying peoples in those areas of low-pop ulation density, the survey was designed to study popu lations living in similar circumstances but that were more accessible logistically. Results from these areas then were used to estimate trachomatous rates for those peoples liv ing in the low-population density areas that were not sur veyed. Areas selected for the survey were: Baringo, Ka jiado, Kakamega, Kisii, Kwale, Meru, Nyanza, and Nyeri.
Clinical Methods In the first five site surveys, examinations were conducted at local facilities (schools, churches, health facilities) con-
Table 1. Characteristics of Regional Survey Areas Sample Size
Estimate% of Rural Population
Meru(Meru)
1142
16.1
Hills
Nyeri (Kikuyu)
1825
20.4
Mt. Kenya highlands
Kwale (Mijikenda)
1342
6.8
Kakamega (Luhya)
1651
23.6
Plateau
Nyanza (Luo)
1807
11.5
Lake Victoria basin
Kisii (Kisii)
1753
6.2
District (tribal group)
Baringo (Tugen)
702
Geography (altitude)
Coastal plain
Southwestern plateau Tugen hills
6.7
(Tugen)
480
(Njemps)
581
Coastal plain
(Pokot)
596
Edge of desert
Kajiado (Maasai)
1924
8.7
Plateau
Grassland plateau
Climate (rainfall)* Temperate to hot (moderate) Cool, temperate (high) Hot, humid, (moderate) Temperate to hot (moderate) Hot, humid (moderate) Temperate (high) Temperate to hot (moderate) Hot (moderate to low) Hot (arid, but near irrigation) Very hot (arid, semi-desert) Hot (arid, semi desert)
Economy* Coffee, tea, "chat" cultivation; subsistence farming Coffee and tea farming; subsistence farming Fishing; tourism; subsistence farming Cotton farming; some coffee farming; subsistence farming Cotton farming; fishing; subsistence farming Tea and coffee farming; subsistence farming Coffee farming; subsistence farming Subsistence farming; herding livestock Herding livestock; some subsistence farming Herding livestock Herding livestock
• Specific rainfall data and regional economic indicators data are not available from published sources or from the government of Kenya.
477
Ophthalmology
Volume 102, Number 3, March 1995
Survey Prevalence of Trachoma Area 62.6% 1. Meru 10.7% 2. Nyeri 0.8% 3. Kwale 0.1% 4. Kakamega 0.2% 5. Nyanza 0.0% 6. Kisii 29.2% 7. Baringo 56.8% 8. Kajiado
Figure 2. Survey areas, pop ulation density, and preva lence of active trachoma.
Population (per sq. km.)
D
venient to the residents of the clusters. Fewer randomly selected survey members were lost to the study when in dividuals were examined at home; survey members were sometimes reluctant to travel any distance away from their households to be examined. To simplify logistics and im prove coverage, survey members were examined in their own households in the remaining three surveys. Each sur vey team consisted ofa minim urn ofan ophthalmologist, an ophthalmic clinical officer or ophthalmic assistant, a registrar, and a locally hired clerk who was familiar with the survey area and local language. Each survey subject was registered, and visual acuity was measured in each eye. An ophthalmic history was taken by the clinical of ficer. All survey members then were examined by the ophthalmologist. Ophthalmologists were members of either the staff of the Kenya Rural Blindness Prevention Project or the International Eye Foundation, or they worked for the Kenya Ministry of Health. There were nine separate examiners over the course of the eight surveys, but the three resident Kenya Rural Blindness Prevention Proj ect ophthalmologists examined 69% of the total number of study subjects. Each examination was carried out according to a structured clinical protocol, with a focal handlight, X4 loupes, and an ophthalmoscope. The examination consisted of a corneal evaluation and examination of the bulbar and tarsal conjunctiva. Every survey subject was examined for evidence of current trachomatous infection or for evidence of previous infection (scars on the tarsal conjunctiva, pannus, Herbert pits). Tra choma was scored according to recommendations of the World Health Organization 17 current at the time of the surveys using the following indicators and cri teria (Table 2).
478
=under20
[[ill
=21 to 149
~
= 150 and over
Methods of Analysis Because of the great sensitivity of estimates of the prev alence and causes of visual loss to the distribution of age in a population, a direct standardization procedure was carried out to minimize the effects of random differences in the age distribution in each of the eight regional survey samples. Each sample population was adjusted to the na tional population age distribution from the 1979 census by weighing age categories according to whether they were over- or under-represented relative to the standard. This procedure allows more appropriate regional comparisons and ensures that national estimates are less influenced by sampling differences. Because the survey districts and the locations within them were not selected randomly, there is no unbiased procedure for constructing true population estimates of ocular conditions. Within each region, it was assumed that the particular locations from which the sample was randomly drawn were representative of a broader popu lation group and that estimates of rates from the survey sample would apply to the underlying regional population within reasonable bounds of error. National estimates were computed by attributing to each regional survey a certain proportion of the national rural population based on similarity in tribal composition, geography, and climate (Table 1). Observations within each survey were appropriately weighted so that the na tional estimates reported reflect these attributed propor tions. For these reasons, and because the sample was drawn in clusters, there is no unbiased method to calculate sam pling errors or confidence intervals for the prevalence es timates. No standard errors are reported, although it would be possible to compute rough estimates based on as
Schwab et al · Trachoma in Rural Kenya Table 2. Criteria for Grading Trachoma in Rural Kenya Survey Criteria
Indicator Intensity of Inflammation Mild Moderate Severe Inactive
At least 5 follicles in superior tarsal plate zones 1 and 2, but fewer than 5 mature follicles 5 or more mature follicles in zones 1 and 2 with or without papillary hypertrophy Marked papillary hypertrophy with or without follicles Lid scarring, Herbert's pits, pannus without any signs of current inflammation
Type and Severity of lid Scarring Mild Moderate Entropion Trichiasis
Fine linear scars in zone 2 Sheet scarring in zone 2 with no lid deformity Deformity of lid with no lashes abrading conjunctiva or cornea Lid deformity and lashes abrading conjunctiva or cornea
Type and Severity of Corneal Lesion Leukoma 1 Leukoma 2 Leukoma 3 Keratitis
Corneal scar with no effect on visual acuity Scar causing slight vision loss Scar across the visual axis Punctate corneal lesions due to trachoma
sumptions of simple random sampling from the infor mation on sample size presented and to inflate them by an appropriate factor (perhaps 2-4 times 18) to account for the nonrandomized design and clustering.
Results An estimated 19% of rural Kenyans have active trachoma, and 12% not currently infected have sequelae of previous infections. Of persons who have clinical signs ofexposure to current or previous trachoma, approximately one in six have deformed eyelids as a result, placing them at high risk of trachoma-induced visual loss, whereas I in 44 have already had some visual loss due directly to the disease. The occurrence of trachoma in rural Kenya is by no means a uniform phenomenon (Fig 3). Overall prevalence oftrachoma inflammation was 63% and 57% in the two most severely affected survey regions, 29% and 11% in
Table 3. Prevalence of Trachoma by Survey Area*
Prevalence
100%.----------------------------------------, 80% 60% 40% 20% 0%
0-4
two others, and less than 1% in the remaining four. Tra choma was most likely to be endemic in arid areas oflow population density, including Baringo, Kajiado, and Meru. Trachoma was less prevalent in areas of denser settlement. Both cash crop and subsistence agriculture is practiced in those areas ofdenser settlement where rainfall is higher and sufficient to sustain it. It also was found to be present but less prevalent in areas of dense human settlement or with higher annual rainfall. Table 3 displays the wide variation of prevalence rates by regional survey site. Figure 3 presents the distribution of trachoma grades in these areas by age cohort in regions where prevalence of active disease is greater than 1%. Overall rates of in flammation peak in the first decade (in fact, in the first 3 years) and then decline rapidly to approximately 20 years ofage and continue to decline more gradually throughout adulthood to old age. In each age group, approximately 50% ofall trachomatous inflammation is classified as mild,
10-14 20-24 30-34 40-44 50-54 60-64 70-74 5-9 15-19 25-29 35-39 45-49 55-59 65-69 75+
Age Group Figure 3. Trachoma prevalence by age group in rural Kenya. Solid bars = mildly active trachoma; shaded bars = moderately to severely active trachoma; open bars = signs of past trachoma.
Survey Area Meru Nyeri Kwale Kakamega Nyanza Kisii Baringo Kajiado
Prevalence of Trachoma (%)
62.6 10.7 0.8 0.1 0.2 0.0 29.2 56.8
* These data correlate by number with survey sites displayed in Fig ure L
479
Ophthalmology
Prevalence
Volume 102, Number 3, March 1995
FEMALES
MALES Prevalence
100%.-----------------------.
Figure 4. Sequelae of tra choma by sex and age group in rural Kenya. Solid bars = corneal scarring; shaded bars = potentially blinding lid de formity; open bars = lid scar ring of any severity.
80%
80%
60%
60%
40%
40%
20%
20%
0%
.__~--
0-9 20-29 40-49 60.09 10-19 30-39 50-59 70 +
0-9 20-29 40-49 60.09 10-19 30-39 50-59 70 +
Age Group
Age Group
and 50% is moderate to severe. The age-specific prevalence among those with any indication of either present or pre vious trachoma reaches 80% to 90% by the fifth decade. Slightly to moderately higher prevalence of infection is seen in females than in males in every group, with the relative risk varying from 1.0 to 2.0. This increased ex posure in females is responsible for higher rates of tra chomatous lid scarring in women, especially of dense scarring causing lid deformities (Fig 4). In addition, women show higher rates of related visually disabling le sions such as corneal leukoma and punctate keratitis. Within the geographically circumscribed and relatively homogeneous areas of each survey, the prevalence of in flammatory trachoma and trachomatous lid scarring is widely variable. This is demonstrated by the within-cluster rates in each region (Fig 5). Although variations are wide, the within-cluster rates tend, nonetheless, to fall within characteristic ranges in each survey area, except in Bar ingo, which was the only survey planned to take into ac count wide ecologic and cultural variability. Rates ofcurrent infection within sample clusters tended to vary with presence of trachoma lid scarring (r = 0.88, P < 0.05), which indicates a continuity between historic and current factors relating to trachoma. The relation is not completely consistent, and in two areas (Nyeri and Nyanza) the average rate oflid scarring is relatively greater than might be expected, given the current prevalence of active disease. Severe lid deformities (entropion with tri chiasis), which greatly increase the risk of visual loss, are the result of repeated re-infection. The cluster-specific prevalence of these severe deformities in all endemic areas increases significantly in proportion to the rate of active inflammation (r = 0.79, P < 0.05). Just as overall prevalence rates vary within and across survey regions, the age-related profile of the disease also varies (Fig 6). In Kajiado, Meru, and Nyeri, the peak prevalence of inflammatory trachoma occurs in children younger than 5 years of age with consistently lower rates
480
100% . - - - - - - - - - - - - - - . ,
in older age groups. In Baringo, there are local prevalence peaks in children younger than 5 years of age and lower rates in older children. Rates of trachomatous infection in Baringo tend to be higher in adults, and those individ uals who are older than 40 years of age are the most se verely affected. These findings parallel those in a recent village trachoma study in Tanzania. 19 The age-associated pattern of trachomatous infection varies within areas of relatively uniform cultural practices and environmental conditions. In three sampling locations in Kajiado, children younger than 5 years of age had high rates of infection, and rates were relatively constant at much lower levels in older age groups. In the other three sampling areas, rates of infection were very high among children younger than 5 years of age, remained at much higher levels among older children and adolescents, and declined slowly with age. Prevalence 100%~------------------------------------.
Meru
Kwale (23)
(30)
Nyeri (30)
Nyanza (30)
Kakamega (29)
Baringo (39) Kisii
(30)
Kajiado (34)
Survey Area (Number of Clusters)
Figure 5. Variation in prevalence of active trachoma in 60-person sample clusters.
Schwab et al · Trachoma in Rural Kenya
NYERI
Prevalence
MERU
Prevalence
10~kr---------------------------~
60% 40% 20%'-····.,.~·-·········
Oo/oL_~~~==~~~~~==:::=:J 0-2
3-5
S.B
12-14 18-20 31-40 51-60 9-11 15-17 21-30 41-50 61+
S.B
0%0-2 3-5
Age Group
Age Group
BARINGO
Prevalence
KAJIADO
100% ,----------------------------~ 80%
12-14 18-20 31-40 51-60 9-11 15-17 21-30 41-50 61+
......................-··
Figure 6. Prevalence of active trachoma in endemic survey areas by age group.
Prevalence
100% r---------------------------..,
·.............
-1----------------------,~~---1
60o/o f-->·o::···-=··:-.:------:=---4----------,;::,-----l ········~······· ~
.......
--
... ,.
40%-1---------------------~--~---1
./ 20o/o t"-o;;;;;:;:'~~.,=-=---::/,._-~,.,-------------j 0%
-·-·-·---~-·-·-·-·-·-·-·-··
0-2
Pokot
3-5
S.B
12-14 18-20 31-40 51-60 9-11 15-17 21-30 41-50 61+
Age Group Plains Tugen Njemps Hill Tugen
Discussion Trachoma remains a major cause of preventable blindness in Africa and in the world. Its endemic status and severity are in part determined by level of economic development, access to clean water, availability of adequate sanitation, and access to preventive and curative health services. Be cause trachoma is a communicable disease of relatively low infectivity, it requires high frequency and duration of exposure to persist in a population. Transmission of the disease at the community level can be interrupted by improved personal hygiene and by provision of an ade quate supply of clean water. Prevalence and severity of infection can decrease markedly as a community under goes socioeconomic development. 20 The results of the Kenya blindness surveys support these findings. Kenyans who live in the arid and semi-arid areas of the country primarily are involved in pastoral activity. These groups are represented in the surveys by the Maasai and Kalenjin nomads and by the semi-nomads ofBaringo district. Most of the people in these areas live outside Kenya's money economy at a lower standard of living than the more settled population groups. It is among these nomadic groups that the prevalence of active trachoma is the highest. An exception is Meru district, where trachoma prev alence was the highest of any district surveyed, despite the fact that rainfall is moderate and considerable cash is generated by the cultivation of tea, coffee, and "chat," a leaf which is chewed as a stimulant, for which there is a
12-14 18-20 31-40 51-60 9-11 15-17 21-30 41-50 61+
Age Group
regional cash market in Kenya and neighboring Somalia. Despite its moderate rainfall, the pumaceous basaltic soil in Meru permits such rapid runoff of rain from this high land area that water remains relatively inaccessible. Al though the people of Meru are not nomadic, the district itself is nevertheless remote and underserved. There are fewer community development projects that could elevate the standard ofliving in Meru District than in other survey districts with similar settled populations. The Meru area is more similar to the arid districts in both ecologic profile and trachoma prevalence because of these factors. It is instructive to contrast the relative prevalence rates of active versus cicatricial trachoma among districts. Lid scarring from inactive trachoma was found to be more prevalent than active disease in the Nyanza and Nyeri districts and to a lesser degree in the Kisii district. In these areas, commercial agriculture has been developed to a greater extent due to favorable rainfall, and, as a conse quence, the standard of living is higher than in districts where active trachoma was found to be more prevalent. The presence of healed scars indicates that trachoma was prevalent within the lifetimes of older people, but eco nomic development and access to improved health ser vices have caused the disease to disappear as a serious public health problem.
Summary Inflammatory trachoma remains a serious problem mainly in areas of rural Kenya where arid climatic con
481
Ophthalmology
Volume 102, Number 3, March 1995
ditions and poor access to water exist and where nomadic lifestyles prevail. In some areas where inflammatory tra choma does not exist, there is evidence of historic disease in older age cohorts. In those areas, economic develop ment, improvements in sanitation, greater access to clean water, and access to health services appear to have con tributed to the elimination of trachoma. From the public health planning perspective, emphasis on eye health ser vices in those areas must be on prevention of blindness due to the sequelae of trachoma rather than elimination of infection.
Appendix Members of the Study Survey Group H. Adala, N. Bakker, H. Beekhuis, G. Bisley, N. Dekkers,
I. Hoshiwara, H. Mandalia, P. Mandalia, R. Meaders,
J. 0. Ochola, and V. Sheffield.
References 1. Sukwa TY, Ngalande TC, Mwandu DH, et al. Prevalence and distribution oftrachoma in the Luapula Valley, Zambia. East Afr Med J 1992;69:34-6. 2. Katz J, Zeger SL, Tielsch JM. Village and household clus tering of xerophthalmia and trachoma. Int J Epidemiol 1988;17:865-9. 3. Loewenthal R, Pe'er J. A prevalence survey of ophthalmic diseases among the Turkana tribe in north-west Kenya. Br J Ophthalmol 1990;74:84-8. 4. De Sole G. Impact of cattle on the prevalence and severity of trachoma. Br J Ophthalmol 1987;71 :873-6. 5. Kabo AM, Di M. La trachome dans le Departement de Zinder (Niger). Rev Int Trach Pathol Ocul Trop Subtrop Sante Publique 1989;66:43-54. 6. Huguet P. Le trachome au Niger: (resultats d'une enquete sondage de Tahoua). Rev Int Trach Pathol Ocul Trop Sub trop Sante Publique 1989;66:31-41.
482
7. Tielsch JM, West KP Jr, Katz J, et al. The epidemiology of trachoma in southern Malawi. Am J Trop Med Hyg 1988;38:393-9. 8. Faal H, Minassian D, Sowa S, Foster A. National survey of blindness and low vision in The Gambia: results. Br J Ophthalmol1989;73:82-7. 9. Arbuckle RD. The International Eye Foundation/Kenya Rural Blindness Prevention Project. Soc Sci Med 1983;17: 1789-92. 10. Bailey R, Osmond C, Mabey DC, et al. Analysis ofhousehold distribution oftrachoma in a Gambian village using a Monte Carlo simulation procedure. Int J Epidemiol 1989; 18:944 51. II. Prost A, Negrel AD. Water, trachoma and conjunctivitis. Bull World Health Organ 1989;67:9-18. 12. Taylor HR, West SK, Mmbaga BBO, et al. Hygiene factors and increased risk of trachoma in central Tanzania. Arch Ophthalmol 1989;107:1821-5. 13. Whitfield R, Schwab L, Ross-Degnan D, et al. Blindness and eye disease in Kenya: ocular status survey results from the Kenya Rural Blindness Prevention Project. Br J Ophthalmol 1990;74:333-40. 14. Whitfield R Jr, Schwab L, Bakker NJA, et al. Cataract and corneal opacity are the main causes ofblindness in the Sam bum tribe of Kenya. Ophthalmic Surg 1983; 14:139-44. 15. Ross-Degnan D, Schwab L, Mburu FM. Field methodology for ocular surveys in rural Africa. Soc Sci Med 1983;17: 1793-6. 16. Henderson RH, Sundaresan T. Cluster sampling to assess immunization coverage: a review of experience with a sim plified sampling method. Bull World Health Organ 1982;60: 253-60. 17. Dawson CR, Jones BR, Tarizzo ML. Guide to Trachoma Control in Programmes for the Prevention of Blindness. Geneva: World Health Organization, 1981. 18. Tabbara KF, Ross-Degnan D. Blindness in Saudi Arabia. JAMA 1986;255:3378-84. 19. Turner VM, West SK, Munoz B, et al. Risk factors for tri chiasis in women in Kongwa, Tanzania: a case-control study. Int J Epidemiol 1993;22:341-7. 20. Meredith SJ, Peach HG, Devanesen D. Trachoma in the Northern Territory of Australia, 1940-1986. Med J Aust 1989;151:190, 192, 194-9.