Trachoma is considered the leading infectious cause of blindness worldwide (Hu et al. 2010) and is caused by serotypes A, B, Ba and C of the bacterium Chlamydia trachomatis. These ocular serotypes cause a follicular keratoconjunctivitis, with the follicles being found on the tarsus of the eyelid (trachoma follicular, TF). Chronic, repeated infection with the ocular serotypes can cause long-term scarring and damage to the eyelid, resulting in in-turning of the eyelashes (trachomatous trichiasis, TT), which can damage the cornea. This can eventually lead to corneal scarring and blindness. In 2002, it was estimated that 1.3 million people had lost their sight as a result of repeated C. trachomatis infection (Resnikoff et al. 2004). In 2009, the number of people with active trachoma was estimated to be approximately 40.6 million and 8.2 million were thought to have trichiasis (Mariotti et al. 2009). In 1997, the WHO began the Global Elimination of Blinding Trachoma by 2020 (GET 2020) Programme, which hopes to eliminate trachoma as a public health problem by 2020. Elimination of trachoma in a country is defined as a prevalence of TF of less than 5% in children aged 1–9 years and fewer than one of 1000 cases of trichiasis throughout all age groups nationally (Report of the 16th Meeting of the WHO Alliance for the Elimination of Blinding Trachoma by 2020). The GET 2020 programme supports implementation of the SAFE strategy of trachoma control: Surgical intervention for trichiasis, Antibiotics to treat active trachoma, Facial cleanliness and Environmental improvements to reduce transmission. Each component of the SAFE strategy is important for the successful elimination of trachoma (Mariotti et al. 2009).
The Gambia was considered to have a low trachoma prevalence [prevalence of TF in two formerly endemic regions in 2006 in children aged 1–9 years was 10.74% (Harding-Esch et al. 2009)] even before the country was mass-treated with azithromycin for trachoma between 2008 and 2010. This 3-year programme consisted of mass azithromycin treatment in 11 districts by the National Eye Health Programme (NEHP, formerly the National Eye Care Programme). In 2010, a 3-year surveillance programme started to determine the prevalence of TF and TT nationally. The surveillance programme excludes urban areas as advised by WHO because the population is thought to be at a much lower risk and these areas are harder to survey. The rationale behind this exclusion is the assumption that urban areas have better access to facilities and services (Solomon et al. 2002). However, The Gambian community of Brikama is a rapidly urbanising area that experiences a great deal of migration, and is frequently used as a transit point. With rapid urbanisation come social problems, such as sanitation and waste disposal, which are known risk factors for the transmission of trachoma (Kuper et al. 2003; Emerson et al. 2006). In Brikama, it is possible that the population is outstripping the available resources thereby increasing the risk factors for trachoma. Constant migratory movement may also be a major risk factor for trachoma, and contact with untreated populations may result in the re-emergence of infection in mass-treated communities (Burton et al. 2005; Shah et al. ). Migration of infected persons into Brikama may cause infection to be introduced into the community. The Gambian population as a whole is highly migratory; this includes close contact with the neighbouring countries of Senegal and Guinea-Bissau which have significantly higher prevalences of trachoma. The prevalence of TF in children aged 1–9 years in villages in the Casamance region of Senegal on The Gambian border ranged from 0% to 15.2% (personal communication, Emma Harding-Esch), and a cross-sectional survey of children aged 1–9 years in the region of Thiès in Senegal found a prevalence of TF of 53.6% (Moalic et al. 2000). The national prevalence of TF in children aged 1–9 years in Guinea-Bissau was 20.2% (Report on the eleventh meeting of the WHO alliance for the Global Elimination of Blinding Trachoma 2007).
This study aimed to determine the prevalence of TF and trichiasis in Brikama to decide if trachoma has been eliminated as a public health problem, or whether further surveillance or intervention is necessary. Risk factor data were also collected to identify potential risk factors for TF in this area.
Ethical approval was obtained from The Gambia Government/Medical Research Council Unit, The Gambia Joint Ethics Committee and the Internal Ethical Review Committee at the University of Birmingham, UK. Written (thumbprint or signature) informed consent was obtained from the household head or the acting household head if the former was not present for eye examination and questionnaire completion.
A population-based cross-sectional prevalence survey was conducted in January–April 2012. WHO sample size calculation was employed (Solomon et al. 2002), using 95% confidence intervals, a design effect of 4 to account for trachoma clustering, an assumed prevalence of TF of 3% and an absolute precision of ±2%. This would allow the survey to determine whether the prevalence of TF in children aged 1–9 in Brikama was less than 5%, which is the level at which elimination of active trachoma as a public health problem is considered to have been achieved.
Brikama is divided into a total of 97 enumeration areas (EAs) by The Gambia's census department, which correspond to different geographical areas of Brikama. From these, 12 EAs were randomly selected. Household head lists were made for the selected EAs, and from each list, 27 households were randomly selected. The selection was made by generating random numbers in Excel (Microsoft Office v. 2003) and then dividing this number by the reciprocal of the number of households in the EA rounding all fractions upwards. Households were sequentially selected from the top of this list, excluding duplicates, until 27 households had been selected. This made up the sample size on the assumption of 3.5 children aged 1–9 years per household. Three reserve households for each EA were further selected in the event of households being unable or unwilling to participate.
Each household was visited at home. The de facto (those who had slept in the household the night before) population was enumerated. All members of the household were offered an ocular examination by an MRC fieldworker who was an experienced clinical trachoma grader. One grader was used for all ocular examinations to ensure consistency. Both eyes were examined using a 2.5× loupe and torchlight, and information on facial cleanliness was collected for each participant. Grading followed the WHO simplified grading system (Thylefors et al. 2010).
Household-level risk factor data were collected via an orally delivered questionnaire. The same fieldworker was used each time to ensure consistency. The questions related to the family's socio-economic status, water and latrine access and whether they had recently had visitors or had recently travelled.
Once data collection was complete, the field team returned and treated all cases of TF and their household contacts with azithromycin (donated by Pfizer via the International Trachoma Initiative), as is NEHP policy. All cases of trichiasis were referred to the NEHP for further treatment. Any other ocular problems were managed according to NEHP guidelines.
All data were entered into a Microsoft Access database (Microsoft Office v. 2003). Data cleaning and analysis were performed in SPSS (SPSS v.19). The overall prevalence of TF in children aged 1–9 years and the prevalence of TT in adults ≥15 years were calculated as a percentage with the corresponding 95% confidence intervals (CI). The prevalence of TF in children aged 1–9 years was also calculated for each EA. Univariate logistic regression was used to look for associations between the presence of TF and risk factor data.
Any significant variables (P ≤ 0.1) were then put into a multivariate logistic regression using a forward stepwise approach to adjust for confounding factors. After the a priori addition of age and sex into the model, the covariates added were having a dirty face (defined here as having dirt on the face with or without ocular or nasal discharge), ocular and nasal discharge, how long the family had lived in Brikama (categorised as <1 year, <5 years or ≥5 years), how many buckets of water were used per day (categorised as <1, 1–2, 3–4 or >4), how many times the child's face was washed per day (categorised as 0,1 or >1) and where the family disposed of their rubbish (categorised as inside, just outside or in a central area). To create a final model, significant variables from the multivariate analysis were examined to detect the presence of multicollinearity between variables (if one variable explained another in the model). These variables and any variables which were non-significant in the multivariable analysis were then excluded, and an enter method was used to create a final model from the remaining variables.
Within 324 households, 652 children aged 1–9 years and 1080 adults aged ≥15 years were examined. 16 children aged 1–9 years were enumerated but not examined, as they were either unavailable on both visits by the field team or were unwilling to participate. Of these 16, nine were in the 6–9 years age category, and 62% were female.
The overall prevalence of TF in children aged 1–9 years was 3.8% (25/652, 95% CI 2.5–5.6). The highest prevalence of TF was found in children aged 6–9 years (5.4%) followed by ages 3–5 (4.1%) and ages 1–2 (1.6%; Table 1). Bilateral TF was found in 18 cases (72%). There were no cases of TI in children aged 1–9 years.
Table 1. Univariate analysisa – child-level variables
|6–9||221||12 (5.4)||1.00|| || |
|Female||337||10 (3.0)||1.00|| || |
|Yes||61||13 (21.3)||0.08||0.03–0.18|| <0.001 |
|No||591||12 (2.0)||1.00|| || |
|Yes||21||4 (19.0)||1.00|| || |
|No||631||21 (3.3)||0.15||0.45–0.47|| 0.001 |
|Yes||128||13 (10.2)||1.00|| || |
|No||524||12 (2.3)||0.21||0.09–0.47|| <0.001 |
|Flies on face|
|Yes||2||0 (0.0)||1.00|| || |
The prevalence of TF varied greatly by EA, ranging from 0% to 8.4%. Three EAs had a prevalence above 5% (the level of elimination): 8.4%, 7.7% and 6%. Three EAs had no cases of active trachoma (TF or TI).
Five cases of trichiasis were found in 1080 adults, resulting in an overall trichiasis prevalence of 0.46% (95% CI 0.17–1.14).
In Brikama, 98% (316/321) of families had access to a latrine; 93% (300/324) were less than 30 min away from their nearest water source; and 99% (321/324) were less than 1 km from their water source. The proportion of children aged 1–9 years who went to school was 78.8% (508/644).
The proportion of children whose families did not originate in Brikama was 67.6% (436/645), of which 17.3% (74/429) were from outside The Gambia.
The proportions of children aged 1–9 years with a dirty face, ocular discharge, or nasal discharge were 9.4% (61/652), 3.2% (21/652) and 19.6% (128/652), respectively (Table 1). Only 0.3% (2/652) of children had flies on their face at the time of examination.
In univariate analysis, there was an increased risk of TF in children aged 1–9 years who had a dirty face, ocular or nasal discharge (P < 0.001) as well as an increased risk if the children had lived in Brikama for <1 year or were washed with 3–4 buckets of water per day as opposed to >4 (P < 0.05). There was also a possible weak unadjusted effect of the children's faces being washed once as opposed to more than once (P = 0.068) and the family disposing of rubbish inside the house (P = 0.058) instead of outside the house. When age was modelled as a categorical variable, there was a weak unadjusted protective effect of being aged 1–2 as opposed to being older (P = 0.054; Tables 1 and 2).
Table 2. Univariate analysisa – household-level variables (a) part 1 and (b) part 2
| (a) |
|Other||60||1 (1.7)||1.00|| || |
|Children at school|
|No||136||8 (5.9)||1.00|| || |
|No||395||16 (4.1)||1.00|| || |
|No||570||22 (3.9)||1.00|| || |
|Born in Brikama|
|No||436||15 (3.4)||1.00|| || |
|How long if no|
|<1 year||15||2 (13.3)||6.17||1.19–32.01|| 0.03 |
|<5 years||42||2 (4.8)||2.01||0.41–9.78||0.39|
|>5 years||329||8 (2.4)||1.00|| || |
|Inside Gambia||355||15 (4.2)||0.00||0.00||0.99|
|Outside Gambia||74||0 (0.0)||1.00|| || |
| (b) |
|No||7||0 (0.0)||1.00|| || |
|Latrine in HHb|
|In another HH||37||2 (5.4)||1.00|| || |
|Open well||208||8 (3.8)||0.98||0.42–2.31||0.97|
|Well with pump||3||0 (0.0)||0.00||0.00||0.99|
|Tap||434||17 (3.9)||1.00|| || |
|Distance to water|
|<1 km||645||25 (3.9)||0.00||0.00||0.99|
|>1 km||7||0 (0.0)||1.00|| || |
|Time to water|
|<30 min||561||22 (3.9)||1.19||0.35–4.08||0.77|
|>30 min||91||3 (3.3)||1.00|| || |
|3–4||162||11 (6.8)||2.52||1.10–5.74|| 0.03 |
|>4||462||13 (2.8)||1.00|| || |
|Freq. children washed|
|1||53||5 (9.4)||2.61||0.93–7.30|| 0.07 |
|>1||495||19 (3.8)||1.00|| || |
|Inside house||375||18 (4.8)||7.06||0.93–53.38|| 0.06 |
|Just outside||120||4 (3.3)||4.83||0.53–43.79||0.16|
|Central area||141||1 (0.7)||1.00|| || |
The initial multivariate analysis found that being male, having a dirty face and the child's family having lived in Brikama for <1 year were associated with a significantly increased risk of having TF. However, it was found that there was great collinearity (the factors explained the same variation in the results) between being male and having a dirty face, ocular or nasal discharge. As a result, sex was removed from the model, and the final multivariable regression model demonstrated that having a dirty face (P < 0.01, OR = 9.23, 95% CI 1.97–43.23), having nasal discharge (P = 0.039, OR = 5.11, 95% CI 1.08–24.10) and having lived in Brikama for <1 year (P = 0.047, OR = 7.78, 95% CI 1.03–59.03) remained associated with an increased risk of TF (Table 3).
Table 3. Final risk factor model
|6–9||221||12 (5.4)||1.00|| || ||1.0||–||–|
|Yes||61||13 (21.3)||0.08||0.03–0.18||0.0001||9.23||1.97–43.23|| 0.005 |
|No||591||12 (2.0)||1.00|| || ||1.0||–||–|
|Yes||21||4 (19.0)||1.00|| || ||6.82||0.47–99.37||0.16|
|No||128||13 (10.2)||1.00|| || ||1.0||–||–|
|Yes||524||12 (2.3)||0.21||0.09–0.47||0.0001||5.11||1.08–24.10|| 0.04 |
|How long in Brikama?|
|<1 year||15||2 (13.3)||6.17||1.191–32.01||0.03||7.78||1.03–59.03|| 0.047 |
|<5 years||42||2 (4.8)||2.01||0.412–9.78||0.39||4.69||0.72–30.62||0.12|
|≥5 years||329||8 (2.4)||1.00|| || ||1.0||–||–|
Population-based prevalence surveys using cluster random sampling (such as this one) remain the ‘gold’ standard for assessing and monitoring the trachoma status of a community (Ngondi et al. 2009), and are essential in the planning of control measures and programmes. As expected, the overall prevalence of TF in Brikama was less than the level at which elimination is considered to have been achieved by the WHO. However, as there was considerable heterogeneity in the results, this does not exclude the existence of TF hotspots with a prevalence of >10% in Brikama. Trachoma is known to cluster in areas, so these results follow the accepted epidemiological manifestation of the disease (Bailey et al. 1989, 1994; Mabey et al. 1992; Burton et al. 2003). However, it was not expected that areas in an urban environment would have a prevalence above 5%. Although in Brikama EA prevalence did not reach 10% and therefore is below the threshold for mass drug administration (Solomon et al. 2002), some areas qualify for implementation of the F and E components of the SAFE strategy (Emerson et al. 2006). These EAs may need to be monitored more closely to ensure that the disease does not spread or increase to any great degree.
There is also the question of whether there is still active infection present in the cases that were found in Brikama, as previous research in The Gambia suggested that C. trachomatis, especially in low prevalence areas, is often not present where there are clinical signs, and there may be a lag time between the clearance of infection and the resolution of clinical signs (Harding-Esch et al. 2009). This has implications for the results of this study because if there is little or no C. trachomatis infection in Brikama, the results of this study may represent this ‘lag’ time. However, it should be noted that mass drug administration with azithromycin in The Gambia took place between 2008 and 2010, and there were three cases of TF in children aged 1–2, which could possibly represent new infection. The clinical grading was performed by an experienced and validated grader, and the use of a single grader ensured consistency. However, other pathogens could be responsible for the TF we observed, for example C. pneumonia or Haemophilius influenzae (Baral et al. 1999; Burton et al. 2011). The trend that is seen with regard to the risk of trachoma decreasing with increased time in Brikama suggests that the infection may not being transmitted and therefore that clinical signs represent past infection or infection acquired in an original community. However, as testing for ocular C. trachomatis does not currently constitute a component of the control programme, decisions can only be made on the levels of follicular trachoma seen.
The association of a dirty face with the presence of TF is a well known and widely accepted risk factor for trachoma (West et al. 2006; Schenneman et al. 2007; Harding-Esch et al. 2008, 2010; Ngondi et al. 2008). The analysis of the presence or absence of a dirty face requires discussion as this relates to the ‘F’ and ‘E’ components of the SAFE strategy to combat trachoma, and the current evidence base for these components is weak (Emerson et al. 2006). It is still unclear whether the presence of discharge associated with trachoma is cause or effect. Discharge may appear because a child has trachoma, or it may be important in the transmission of trachoma through clothing, towels and hands. The presence of discharge could also make it more likely for dirt to stick to a child's face resulting in the risk factor for a dirty face. The presence of ocular discharge was no longer significant after multivariate analysis, and it is likely that this is being explained by nasal discharge or a dirty face, as the three are probably associated. Alternatively, as the study was powered to detect prevalence, it may be underpowered to detect certain risk factors such as these.
The increased risk of having lived in Brikama for less than 1 year being positively associated with TF is a previously undocumented finding, which supports the theory of new migrants bringing infection into Brikama. Although it was not possible (due to power) to do a regression analysis on this population alone, there are many reasons why this population may be at greater risk. They may be a more mobile population that may be less settled in Brikama. This may adversely affect their socio-economic, housing and sanitary situation in a way that the questionnaire was not able to accurately define. This theory is also supported by the trend shown in this finding, as the odds ratio still shows an increased risk if a household had lived there for less than 5 years compared with more than 5 years even though this is not a significant result. This finding could support the theory that population movement may indeed be a risk factor for trachoma (Kuper et al. 2003).
As living in Brikama for less than a year and having a dirty face were risk factors for TF, it was hypothesised that children coming in from outside Brikama might be more likely to have a dirty face and be introducing ‘unhygienic’ practices. However, when this hypothesis was investigated, there was no significant association (P = 0.89). As The Gambia has reduced TF levels in rural areas to elimination thresholds, the difference in the prevalence of TF in children entering Brikama from outside and inside The Gambia was compared. The prevalences were 3.45% (95% CI 2.05–3.45) and 2.92% (95% CI 1.63–4.21), respectively. However, due to the low power of the study, it was not possible to say whether these figures are significantly different.
The potential problem of the study being underpowered to detect risk factors applies to all of the risk factors that were significant after univariate analysis, but not significant after multivariable analysis. However, this is more likely to be due to confounding factors present in the analysis. Although age was not a significant risk factor variable, it is interesting to note that risk appears to increase with age. This is unexpected as it has been noted that younger children are normally at an increased risk in The Gambia (Harding-Esch et al. 2008, 2010). A potential explanation for this is that older children in Brikama may come into more contact with large groups of children where trachoma may be more easily transmitted, such as at school. School attendance may also be a reason for boys having a higher risk of TF than girls. The remaining non-significant risk factors such as how much water a family uses and how many times a child's face is washed probably are true risk factors. However, during multivariable analysis it is likely that they are absorbed into the more general question of whether a child has a dirty face, and the location in which a family disposed of their rubbish could have been confounded by variables, which are proxy measures for how clean a family's house is.
Only 652 children were seen when the sample size required was 1118 children. The sample was calculated on the assumption of 3.5 children per household, which was not accurate for Brikama, as it was based on figures from data collected in rural areas (there were two children aged 1–9 years per household on average in Brikama). Although this does not affect the ability of the study to detect the prevalence of TF in children, it has slightly decreased the accuracy of the prevalence estimate, meaning that we cannot be 95% sure that the prevalence is less than the WHO elimination level of 5%. Any further work in urban areas may need to re-evaluate any assumptions made, which are based on data from rural areas, including the fact that there are more adults in urban areas than rural areas. This may mean that there are more cases of trichiasis in urban areas than are projected from population data.
The sample size calculation was based on a TF prevalence of 3% and an absolute precision of ±2%, and accounted for clustering by employing a design effect of 4. This was to allow us to determine whether the prevalence of TF was below 5% (the WHO target for elimination). As the overall prevalence of TF was 3.8%, the upper confidence interval exceeds 5%, and therefore, it is not possible to conclude that the WHO target for TF elimination has been met. The study was not powered for the risk factor analysis, and therefore, the statistical power for these analyses is at a lower level.
There are several implications from these results. Firstly, the overall prevalence of TF in Brikama does not present a public health issue, and therefore, mass administration of azithromycin is unnecessary. The facial cleanliness and environmental improvement components of the SAFE strategy could be employed, although data show that water and latrine access are good. This suggests that the environmental conditions are not present to sustain trachoma infection if it is introduced from outside. Despite this, it may be prudent to include Brikama in The Gambia's trachoma surveillance programme as it contains areas with a prevalence >5%, the population is continually expanding, and Brikama continues to receive migrants from neighbouring countries. As the evidence points to the fact that people who have more recently arrived in Brikama are at a greater risk of trachoma, it may be worthwhile trying to focus education on facial cleanliness and environmental improvements on this population.
Secondly, and perhaps most importantly, as the prevalence of trichiasis is above the 0.1% level of elimination, there needs to be further active case finding for individuals with trichiasis. At this prevalence, there are a projected 357 cases of trichiasis in Brikama alone (based on the 1993 population data and an average population growth of 10% per year). It is also important to note that people in Brikama all live within close proximity to an eye clinic where surgery for trichiasis is readily available. Four of five trichiasis cases found were unknown to services and this suggests that there are barriers to people accessing this care.
Access to surgery in The Gambia is good but uptake rates for surgery are low (Bowman et al. 2002). There are a number of potential barriers to seeking surgical treatment for trichiasis, such as financial constraints, time constraints, a lack of knowledge of the services available and fear (Bowman et al. 2002; Nagpal et al. 2006). There is only one ophthalmologist available in The Gambia who serves the whole country, but most cases of trichiasis are treated by ophthalmic nurses who have been trained to perform corrective surgery (A. Sillah, personal communication). Community perceptions and understanding of trichiasis are not well documented, but a survey showed that there was poor understanding of the chronic nature of the disease and the connection between childhood infection and trichiasis later in life (Ajewole et al. 2001). To address the problem of trichiasis in Brikama, barriers to treatment in this area need to be identified and removed. In 2010, there were projected to be around 10 000 cases of trichiasis in The Gambia, (Rajak et al. 2010) which has a population of 1.5 million, equating to a trichiasis prevalence that is above the 0.1% level of elimination set by the WHO. Trichiasis is likely to remain the main barrier to The Gambia verifying for elimination of trachoma.
The results of this study have been presented to both the Trachoma Task Force (the body responsible for monitoring trachoma activities/policies) and to representatives from all the health districts of The Gambia at a meeting for the Ministry of Health. They were also discussed at the GET 2020 meeting in April 2013. There have not been any decisions yet as to whether Brikama and urban areas in general should be included in surveillance programmes.
In conclusion, this study has shown that, while not a major public health problem, active trachoma persists in Brikama with an overall TF prevalence of 3.8% (95% CI 2.5–5.6). The prevalence of TF varies by EA, some being above the level of elimination as set by the WHO, which is substantially higher than preliminary data suggests. Children who have dirty faces, nasal discharge and whose families have recently arrived in Brikama are at a higher risk of TF. The prevalence of trichiasis adjusted for the total population of Brikama (not just those >14 years) is 0.26%. In light of these findings, it may be important to include areas such as Brikama in surveillance and education activities to monitor the levels of TF and trichiasis in these communities, especially those areas that receive many immigrants. Active case finding for trichiasis remains the main priority in Brikama for The Gambian NEHP as this is the component of elimination that will prevent The Gambia from verifying for trachoma elimination in the future.