• Loading metrics

Trachoma and Relative Poverty: A Case-Control Study

  • Esmael Habtamu ,

    Affiliations: International Centre for Eye Health, London School of Hygiene and Tropical Medicine, London, United Kingdom, The Carter Center, Addis Ababa, Ethiopia

  • Tariku Wondie,

    Affiliation: The Carter Center, Addis Ababa, Ethiopia

  • Sintayehu Aweke,

    Affiliation: The Carter Center, Addis Ababa, Ethiopia

  • Zerihun Tadesse,

    Affiliation: The Carter Center, Addis Ababa, Ethiopia

  • Mulat Zerihun,

    Affiliation: The Carter Center, Addis Ababa, Ethiopia

  • Zebideru Zewdie,

    Affiliation: Amhara Regional Health Bureau, Bahirdar, Ethiopia

  • Kelly Callahan,

    Affiliation: The Carter Center, Atlanta, Georgia, United States of America

  • Paul M. Emerson,

    Affiliation: International Trachoma Initiative, Atlanta, Georgia, United States of America

  • Hannah Kuper,

    Affiliation: International Centre for Evidence in Disability, London School of Hygiene and Tropical Medicine, London, United Kingdom

  • Robin L. Bailey,

    Affiliation: Clinical Research Department, London School of Hygiene and Tropical Medicine, London, United Kingdom

  • David C. W. Mabey,

    Affiliation: Clinical Research Department, London School of Hygiene and Tropical Medicine, London, United Kingdom

  • Saul N. Rajak,

    Affiliation: International Centre for Eye Health, London School of Hygiene and Tropical Medicine, London, United Kingdom

  • Sarah Polack,

    Affiliation: International Centre for Evidence in Disability, London School of Hygiene and Tropical Medicine, London, United Kingdom

  • Helen A. Weiss,

    Affiliation: Department of Infectious Disease Epidemiology, London School of Hygiene & Tropical Medicine, London, United Kingdom

  • Matthew J. Burton

    Affiliation: International Centre for Eye Health, London School of Hygiene and Tropical Medicine, London, United Kingdom

Trachoma and Relative Poverty: A Case-Control Study

  • Esmael Habtamu, 
  • Tariku Wondie, 
  • Sintayehu Aweke, 
  • Zerihun Tadesse, 
  • Mulat Zerihun, 
  • Zebideru Zewdie, 
  • Kelly Callahan, 
  • Paul M. Emerson, 
  • Hannah Kuper, 
  • Robin L. Bailey



Trachoma is widely considered a disease of poverty. Although there are many epidemiological studies linking trachoma to factors normally associated with poverty, formal quantitative data linking trachoma to household economic poverty within endemic communities is very limited.

Methodology/Principal Findings

Two hundred people with trachomatous trichiasis were recruited through community-based screening in Amhara Region, Ethiopia. These were individually matched by age and gender to 200 controls without trichiasis, selected randomly from the same sub-village as the case. Household economic poverty was measured through (a) A broad set of asset-based wealth indicators and relative household economic poverty determined by principal component analysis (PCA, (b) Self-rated wealth, and (c) Peer-rated wealth. Activity participation data were collected using a modified ‘Stylised Activity List’ developed for the World Bank’s Living Standards Measurement Survey. Trichiasis cases were more likely to belong to poorer households by all measures: asset-based analysis (OR = 2.79; 95%CI: 2.06–3.78; p<0.0001), self-rated wealth (OR, 4.41, 95%CI, 2.75–7.07; p<0.0001) and peer-rated wealth (OR, 8.22, 95% CI, 4.59–14.72; p<0.0001). Cases had less access to latrines (57% v 76.5%, p = <0.0001) and higher person-to-room density (4.0 v 3.31; P = 0.0204) than the controls. Compared to controls, cases were significantly less likely to participate in economically productive activities regardless of visual impairment and other health problems, more likely to report difficulty in performing activities and more likely to receive assistance in performing productive activities.


This study demonstrated a strong association between trachomatous trichiasis and relative poverty, suggesting a bidirectional causative relationship possibly may exist between poverty and trachoma. Implementation of the full SAFE strategy in the context of general improvements might lead to a virtuous cycle of improving health and wealth. Trachoma is a good proxy of inequality within communities and it could be used to target and evaluate interventions for health and poverty alleviation.

Author Summary

Trachoma has long been considered a disease of poverty. However, there is surprisingly little direct data that formally quantifies the relationship between trachoma and economic poverty, and none that specifically focuses on trichiasis. We compared 200 people with trachomatous trichiasis (TT) to 200 people (controls) without the condition, who were matched on age and sex, living in the same community, in Amhara Region, Ethiopia. We measured household relative poverty using three measures: household assets, self-rated wealth and peer-rated wealth. We also measured activity participation. We found TT case households were poorer by all relative economic measures. We found cases less likely to participate in economically productive activities regardless of visual impairment and other health problems, more likely to report difficulty and need assistance performing activities. The results suggest that the causative relationship between poverty and trachoma could possibly be bidirectional: poor households are more affected by trachoma and trichiasis reduces productivity even prior to development of visual impairment, which may exacerbate poverty. Implementation of the SAFE strategy in the context of general socioeconomic improvements might lead to a virtuous cycle of improving health and wealth. Trachoma could be used as proxy of inequality and to target and evaluate interventions for health and poverty alleviation.


Trachoma is leading infectious cause of blindness worldwide [1]. Trachomatous trichiasis (TT) is the late stage consequence of repeated conjunctival Chlamydia trachomatis infection in which eyelashes turn towards the eye, causing pain and eventually irreversible blinding corneal opacification (CO). About 229 million people live in trachoma endemic areas, and approximately 7.3 million have untreated TT [2,3]. More than 2.4 million people are visually impaired from trachoma worldwide, among which between 439,000 and 1.2 million are estimated to be irreversibly blind [2,4]. The WHO recommends the SAFE Strategy for trachoma control [5].This involves Surgery for trichiasis, Antibiotics for infection, Facial cleanliness and Environmental improvements to suppress chlamydial infection and transmission.

Trachoma has long been considered a disease of poverty [6]. It is believed that the decline in trachoma observed in Europe, North America and elsewhere over the last century, in the absence of specific control measures, was largely attributable to general improvements in socio-economic status [7,8]. Trachoma remains prevalent in developing and marginalised communities, particularly in Africa, where crowded living conditions are common and access to clean water, sanitation and health care are often limited [6,8,9]. However, not all people living in such settings acquire active or scarring trachoma. It is possible that, within apparently homogeneous communities, the individuals who are most vulnerable to developing the blinding complications of trachoma are the poorest members of the poorest communities, although this has not been adequately investigated [10]. Moreover, the disability that TT causes may lead to reduced productivity, unemployment and loss of income, putting additional financial pressure on an already strained household [1113]. The effect of trachoma on income may begin prior to the visual impairment, with the pain and the photophobia from trichiasis limiting function [13,14]. Of note, blindness has generally been associated with lower socio-economic status [1517].

In low and middle income countries (LMICs) resources are often shared within households. Therefore, relative wealth or poverty in LMICs needs to be measured at household level, as the economic impact of a medical condition or intervention potentially affects the whole family [18]. In low-income settings estimating income can be difficult, as many people are self-employed and incomes are subject to significant short-term fluctuations [18,19]. In addition, people may earn from sources that they do not wish to disclose. Consumption expenditure data are considered more reliable than income data [16,19]. However, this method is subject to recall bias and requires detailed questionnaires, which are time consuming and costly to administer [19]. An alternative approach is to use a range of asset and housing characteristics as proxy indicators for household wealth and socio-economic status [19,20]. A key advantage of this approach is that it measures the long-term financial status of a household, and is less vulnerable to short-term fluctuations than income and consumption expenditure [19,20]. On the other hand, asset score only measure relative poverty, which may preclude regional or international comparability.

There is surprisingly little direct data that formally quantifies the relationship between trachoma and economic poverty, and none that specifically focuses on the scarring sequelae. The aim of this study was to investigate in detail the relationship between poverty and trachomatous trichiasis through an asset-based analysis, self-rated and peer-rated wealth measures, and participation in productive activities.


Ethics Statement

This study was reviewed and approved by the Food, Medicine and Healthcare Administration and Control Authority of Ethiopia, the National Health Research Ethics Review Committee of the Ethiopian Ministry of Science and Technology, Amhara Regional Health Bureau Research Ethics Review Board Committee, the London School of Hygiene and Tropical Medicine (LSHTM) Ethics Committee, and Emory University Institutional Review Board. Written informed consent in Amharic was obtained prior to enrolment from participants. If the participant was unable to read and write, the information sheet and consent form were read to them and their consent recorded by thumbprint.

Study Design and Participants

This case-control study was nested within a clinical trial of two alternative surgical treatments for trichiasis. From the 1000 trichiasis cases recruited into the trial, every fifth consecutive case was also enrolled into this economic poverty study and matched to a non-trichiasis control. This approach was chosen for logistical and methodological reasons, in order to identify and collect data from controls within the shortest possible time period following case recruitment. Cases were defined as individuals with one or more eyelashes touching the eyeball or with evidence of epilation in either or both eyes in association with tarsal conjunctival scarring. People with trichiasis of other causes, recurrent trichiasis and those under 18 years were excluded. Trichiasis cases were identified mainly through community-based screening. Trichiasis screeners and counsellors (Eye Ambassadors) visited every household in their target village, identified and referred trichiasis cases to health facilities where surgical services were provided. Some individuals self-presented or were referred by local health workers. Recruitment was mainly from three districts of West Gojam Zone, Amhara Region, Ethiopia between February and May 2014. This area has one of the highest burdens of trachoma worldwide [21].

Controls were individuals without clinical evidence or a history of trichiasis (including surgery and epilation), and who came from households without a family member with trichiasis or a history of trichiasis, as we wanted to measure household level relative poverty, which requires comparison of trichiasis case households with households without trichiasis cases. One control was individually matched to each trichiasis case by location, sex and age (+/- two years). The research team visited the sub-village (30–50 households) of the trichiasis case requiring a matched control. A list of all potentially eligible people living in the sub-village of was compiled with the help of the sub-village administrator. One person was randomly selected from this list using a lottery method, given details of the study and invited to participate if eligible. If a selected individual refused or was ineligible, another was randomly selected from the list. When eligible controls were not identified within the sub-village of the case, recruitment was done in the nearest neighbouring sub-village, using the same procedures.

Data Collection

Data on detailed demographic characteristics were collected. Household economic poverty was measured through (a) Asset based wealth indicators, (b) Self-rated wealth, and (c) Peer-rated wealth. Activity participation data was collected using a modified ‘Stylised Activity List’ developed for the World Bank’s Living Standards Measurement Survey [22]. Visual acuity of both cases and controls were measured and cases underwent detailed trachoma examination.

Asset-based wealth inequality indicators.

Data on 60 asset variables were collected. This included (i) housing characteristics and utilities, (ii) ownership of durable assets, and (iii) ownership of agricultural assets. Most data were collected through direct observation. Data on access to water was not collected as it is mainly supplied by government and non-governmental organisations, therefore would not directly reflect the household’s wealth but rather general infrastructure development in the area. Households were asked about their financial savings and whether they had loans from the government at the time of data collection.

Self and peer-rated wealth indexes.

The participants were asked the question: “How well-off do you think your household is in relation to the other households in the village?” They were then asked to choose one of the following options: (1) very poor, (2) poor, (3) average, (4) wealthy or (5) very wealthy. Three members of the village administration team (peers of both the cases and the controls) were then randomly selected and independently asked the question: “How well-off do you think [Name of household head] household is in relation to the other households in the village? for both the case and the control households. They were asked to choose one of the five levels.

Activity participation data.

The ‘Stylised Activity List’ tool contains a list of common activities in different subgroups: household activities, paid work, work for own use, leisure activities and personal activities.[22,23] Participants were asked if they had participated in any of the activities in the subgroups in the last week. If they had undertaken a specific activity in the last week, they were asked the question “How much difficulty did you have in doing [Activity] in the last week? and asked to choose one of the following options: (0) extreme/not able to do, (1) a lot of difficulty, (2) some difficulty, (3) little difficulty, (4) no difficulty; and another question whether they have done the activity (1) fully assisted, (2) with some assistance, (3) with no assistance.

Visual acuity and clinical examination.

Presenting LogMAR (Logarithm of the Minimum Angle of Resolution) visual acuity at two metres was measured using “PeekAcuity” software on a smartphone in a dark room [24]. The ophthalmic examination was conducted in a dark room using a 2.5x binocular magnifying loupe and a bright torch. Clinical signs were graded using the Detailed WHO Follicles Papillae Cicatricae (FPC) Grading System [25].

Sample Size

To detect a difference in asset-based principal component analysis (PCA) similar to that found in the Cataract Impact Study (mean and standard deviation of asset based PCA score in cataract cases and their controls 0.6 and 2.0; and 0.3 and 2.6, respectively) with an alpha of 0.05 and 95% power, at least 346 (173 in each group) participants were required [16]. We recruited 200 trichiasis cases and 200 age, sex and location matched non-trichiasis controls.


Data were double-entered into Access (Microsoft) and transferred to Stata 11 (StataCorp) for analysis. Conditional logistic regression was used to compare basic characteristics of matched cases and controls.

Asset index analysis.

Descriptive and summary statistics of all asset indicators were calculated. A PCA was used to analyse the asset-based wealth or inequality indicator data in order to classify households into different socio-economic levels [19,20,2628]. Variables owned by less than 5% or more than 95% of the participants’ households were excluded from the PCA as they would have the least weight and less value in differentiating socio-economic status or inequality. The PCA was conducted separately to generate a factor score for each of the three subset asset indices: (1) housing and utilities, (2) durable assets and (3) agricultural assets, and for all asset variables combined [19,20,27]. The control households were grouped into quintiles based on the overall asset index socio-economic score (SES). Then the case households were classified based on the “cut points” of the controls’ socio-economic quintiles. We performed matched univariable and multivariable conditional logistic regression analyses to investigate the relationship between asset-based household economic poverty and case-control status. A stratified analysis was performed to test whether the observed association persisted in different groups. Logistic regression adjusted for clustering using robust standard errors was used for stratified analyses of all economic poverty measures by age, sex, marital status and vision, and variables of insufficient frequencies (such as government loan) for matched analysis. Likelihood ratio tests were used to obtain p-values in categorical exposure variables. To test for robustness of the asset index a Spearman rank correlation coefficient was employed to examine whether the three sub-set asset indices produce similar classifications of SES to the overall asset index. To adjust for multiple comparisons, we used the Benjamini and Hochberg method, assuming a false discovery rate (FDR) of 5% [29].

Self and peer-rated wealth indexes.

The wealth scores provided by the three peers were averaged. The association between self and peer-rating of household socio-economic status and case-control status was examined using conditional logistic regression. The self-rated and peer-rated wealth scores were converted into a score out of one hundred, using the formula: ([individual score–lowest possible score]/[Highest possible score—lowest possible score])x100. Lower scores indicate a worse score (0 the lowest possible score) and higher scores indicates better score (100 the highest score) [30]. The mean scores were compared between cases and controls using the Wilcoxon rank-sum test. The correlations of the self-rated wealth, the peer-rated wealth and the asset index based socio-economic classifications of households were compared using Spearman rank correlation coefficient.

Activity and participation data.

Activities undertaken (paid employment and commission work) and not undertaken (talking with friends) by <1% participants were excluded from the analysis. Activities were regrouped into productive household activities, outdoor activities, paid work, agricultural activities and leisure activities. The association between participation in an activity and case-control status were analysed using conditional logistic regression adjusting for self-reported health problems occurring in the last month. Logistic regression adjusted for clustering (using robust standard errors), age, sex and self-reported health problems was used to analyse the difference in activity participation between cases and controls stratified by vision, and to analyse the association between case-control status and difficulty in doing an activity and receiving assistance.

Clinical data.

Presenting visual acuity in the better eye was used in analysis. For visual acuities of counting fingers or less, LogMAR values were attributed as follows: counting fingers, 2.0; hand movements, 2.5; perception of light, 3.0; and no perception of light, 3.5 [31]. The LogMAR visual acuity scores were categorised using the WHO classification: normal vision, ≥6/18; moderate visual impairment, <6/18 to ≥6/60; severe visual impairment, <6/60 to ≥3/60; and blind, <3/60. Corneal opacity grading and trichiasis grading in the more affected eye was undertaken to test their association with household economic poverty among trichiasis cases. Based on their severity, trichiasis cases were categorised into Minor Trichiasis cases with <6 lashes or evidence of epilation in <1/3rd of the lash margin; and Major Trichiasis cases with ≥6 lashes or evidence of epilation in ≥1/3rd of the lash margin.


Demographic and Clinical Characteristics of Participants

Cases and controls were well matched in terms of location, gender and age and had similar levels of literacy, household size and household occupation (Table 1). Compared to the controls, the trichiasis cases were less likely to be married, more likely to be either unemployed or work as daily labourers, less likely to have a family member with formal education and more likely to have experienced a health problem during the last month. As expected, cases were more likely to be visually impaired than the controls (37.0% v 3.0%, respectively; OR = 69.0; 95%CI 9.58–496.82; p<0.0001)

Table 1. Demographic and clinical characteristics of individual participants and their households.

Distribution of Assets

The asset variables used in the PCA are described in Table 2 and their summary statistics are shown in S1 Table. The PCA was based on a combination of 28 asset values. The other 32 measured assets were excluded as they were present in less than 5% or more than 95% of the participants’ households. Households were generally poor. About 67% had a latrine, among which 65% were of the “non-improved” pit latrine type without a concrete slab. About half (54%) had their cattle dwelling within the main house. Ownership of durable assets such as mobile phones and radio was low (<30%). Only 17% of the households had access to electricity. About 12% of the households had taken a government loan. Overall, cases had fewer household and agricultural assets than controls and were more likely to have a government loan (Table 2). There was no difference in the ownership of the house they were living in (92.0% vs 94%, p = 0.22), or access to electricity (18·5% v 16·5%, p = 0·40). Case households had fewer rooms (1.22 vs 1.55, p<0·0001), and had a higher density of persons per room than the controls: 4.0, 95%CI 3.6–4.4 vs 3.3, 95%CI 3.0–3.6 respectively (P = 0.020).

Table 2. Descriptive and summary statistics for all 28 asset variables that were included in the principal component analysis.

Asset Index Factor Scores

The overall asset index accounts for 21% of the total variance (S1 Table). Among the three subset asset indices, the agricultural asset indicators had the highest factor scores and accounted for the highest weights in measuring wealth in this population. In contrast, the housing characteristics and utilities index, except for the number of metal roof sheets, had generally lower factor scores and contributed lower weights in estimating wealth than the other two subset indices. Among all indices, number of oxen and cows owned (0.324), the number of metal roof sheets (0.320) and amount of land owned in hectares (0.319) had the highest weights in estimating wealth. In contrast, access to electricity (-0.096) having cattle dwelling within the main house (-0.024) and having a government loan (-0.038) had negative weights. Fig 1 illustrates the distribution of the subset and overall asset indices, in order to determine whether clumping or truncation were present in this data. Overall, there was evidence of truncation and clumping when the three subset indices (Fig 1A to 1C) are used separately. However, the distribution of the overall combined factor scores was much smoother; and clumping and truncation were not observed (Fig 1D).

Fig 1. Distribution of socio-economic scores for (a) housing characteristics and utilities, (b) durable assets, (c) agricultural assets and (d) all assets combined.

Asset Based Household Economic Poverty and Trichiasis

There was a strong association between being a trichiasis case and asset based household economic poverty: OR = 2.79; 95%CI, 2.06–3.78; p<0.0001 (Table 3). This relationship persisted after adjusting for marital status, and highest family education (OR = 2.78; 95%CI, 2.00–3.87; p<0.0001). For stratified analyses we combined “richest” and “rich” with “middle” because of small numbers, to create a “middle & above” category with three levels of socio-economic status measure to facilitate data modelling. Compared to the controls, trichiasis cases were more likely to be from the poorest (OR = 2.65; 95%CI, 2.05–3.42; p<0.0001) households than from the middle & above households (Table 4). In the stratified analysis, the association between asset based household economic poverty and trichiasis persisted regardless of age, gender, marital status, and in people with normal visual acuity after adjusting for the matching variables and family education (Table 4).

Table 3. Association between household economic poverty and trachomatous trichiasis.

Table 4. The relationship between household economic poverty and trachomatous trichiasis using the asset index, self-rated wealth index and peer-rated wealth index, stratified by age, sex, marital status and vision.

Self and Peer-Rated Wealth Indexes and Trichiasis

On both the self-rated and peer-rated scores, the households of trichiasis cases were rated poorer than controls (Table 3). This association persisted in both self-rated (OR = 3.99; 95%CI, 2.43–6.54; p<0.0001) and peer-rated (OR = 9.10; 95%CI, 4.79–17.27; p<0.0001) wealth measures after adjusting for marital status and highest family education. Compared to the controls, the trichiasis case households were more likely to be rated as poorest and poor rather than middle or affluent by themselves (OR = 3.74; 95%CI, 2.55–5.49; p<0.0001) and their peers (OR = 10.57; 95%CI, 6.42–17.41; p<0.0001) compared to the other households in their villages (Table 4). Using the 0 to 100 scale (poorest to richest), the mean self-rated scores for cases and controls were 34.1 v 49.1 (p<0.0001) and for peer-rated scores they were 27.5 v 50.3 (p<0.0001). The association of lower self-rated and peer-rated wealth with trichiasis persisted regardless of age, gender, marital status, and in people with normal visual acuity after adjusting for the matching variables and family education (Table 4).

Reliability and Correlation of Economic Poverty Measures

The asset based socio-economic classification of households was found to be robust and produced similar ranking of households when the overall index was compared with the different subset indexes; the Spearman rank correlation coefficient ranged between 0.88 and 0.94. A Spearman rank correlation coefficient between asset index and self-rated wealth index, asset index and peer-rated wealth index, and self and peer-rated wealth indexes were 0.58, 0.70 and 0.63, respectively.

Activity Participation and Trichiasis

Trichiasis cases were significantly less likely to participate in household, outdoor, agricultural and leisure activities, even after controlling for the presence of other health problems during the preceding month, (Table 5). However, the trichiasis cases were slightly more likely to participate in daily labouring and self-employment activities such as selling goods. These associations persisted in multivariable analysis after controlling for self reported health problems during the preceding month, except for leisure activities. In stratified analyses by vision, trichiasis cases with normal vision were significantly less likely to participate in processing of agricultural products and in productive outdoor activities such as fetching wood and travelling compared to controls with normal vision (Table 5).

Table 5. Associations between participation in an activity during the last week and case-control status; and stratified analyses by vision.

After adjusting for the matching variables and self reported health problems, trichiasis cases were significantly more likely to report difficulty in performing all productive and leisure activities than the controls: >66% of the cases reported difficulty in all productive activities in contrast to <5% of controls (Table 6). Similarly, trichiasis cases were significantly more likely to report receiving assistance in doing all productive activities compared to controls. In contrast to other activities, higher proportions of trichiasis cases received assistance particularly in agricultural activities such as farming, animal husbandry and processing agricultural products (Table 6).

Table 6. Association between case-control status and having difficulty in doing an activity and receiving assistance to do it among those who have done the activity in the past week.

Factors Associated with Asset Based Household Economic Poverty in Trichiasis Cases

In a univariable analysis (Table 7), being a household head with trichiasis had a strong association with economic poverty (OR = 3.29; 95%CI, 1.89–5.75; p<0.0001) while visual impairment had a borderline association (OR = 1.71; 95%CI, 0.98–2.97; p = 0.058). Not having a marriage partner (OR = 9.41; 95%CI, 4.16–21.31; p<0.0001), no family member with formal education (OR = 4.95; 95%CI, 1.73–14.16; p = 0.0028) and a main family job of daily labouring (OR = 19.64; 95%CI, 2.32–166.49; p = 0.0063) as opposed to farming were independently associated with economic poverty (Table 7). Families in which there were more people of a productive age were less likely to be poor than their counterparts (OR = 0.32; 95%CI, 0.16–0.60; p = 0.0005) (Table 7). In a multivariable analyses, participating in animal husbandry (OR = 0.05; 95%CI, 0.02–0.12; p<0.0001) and agricultural product processing (OR = 0.50; 95%CI, 0.27–0.91; p = 0·024) activities were independently associated with wealthier households while house cleaning (OR = 2.05; 95%CI, 1.03–4.08; p = 0.042) and self employment (OR = 2.77; 95%CI, 1.25–6.18; p = 0.012) activities were associated with poorer households.

Table 7. Univariable and multivariable ordinal logistic regression for household economic poverty among the 200 trichiasis cases only.


Poverty is a complex multidimensional issue that encompasses not only deprivation of material possessions but also wider issues such as nutrition, health and education [32,33]. Many different approaches have been taken to measuring “poverty”, both in absolute and relative terms [34]. In general, these involve a survey methodology to capture estimates of income or consumption and methods that take into account broader issues of health and education such as the Multidimensional Poverty Index [34].

According to the 2011 World Bank estimates, 29.6% (Urban, 25.7%; Rural, 30.4%) of Ethiopians live below the national absolute poverty line (defined as 3781 Birr) and 30.7% live on less than US$1.25 PPP (purchasing power parity) a day [35]. Using asset indicators, the World Bank defines a household as being deprived “when none of these assets are owned by the household: fridge, phone, radio, TV, bicycle, jewelry, or vehicle” [35]. According to these criteria, 53% of rural households in Ethiopia were in deprivation in 2011. However, these are narrowly defined assets and most of these would not be commonly found in a rural Ethiopian community, irrespective to the level of wealth [35].

In this study we compared individuals with trichiasis to matched controls from within the same communities in Amhara Region, Ethiopia using three different measures of relative poverty: Asset Index, Self-Rated Wealth Index and Peer-Rated Wealth Index. These measures allow us to understand whether people with TT were relatively poorer than their neighbours, even within these very poor communities. We performed a PCA of household assets to stratify the participants into economic groupings. The variance explained by the first principle component was similar to the range reported in other similar studies (between 11% and 27%) [19,20,27,36]. The asset index used in this study is probably a reasonable proxy for consumption expenditure as we collected data on a sufficiently broad set of asset indicators that are capable of capturing living standards and wealth inequalities based on local values [37].

Participant and Household Characteristics

The age distribution, gender profile and literacy status of the trichiasis cases in this study were comparable with those reported in our earlier studies in Ethiopia as well as other studies of trichiasis patients elsewhere in Sub-Saharan Africa [31,3840]. This suggests that the results are probably generalizable for this region of Ethiopia at least. The households of trichiasis cases were significantly less well off than controls in terms of ownership of almost all asset indicators measured. Consistent with the literature, trichiasis cases had significantly smaller and more crowded households [6,41]. Cases had less latrine access and more kept their cattle within the house, which is consistent with observations that active trachoma is associated with poor sanitation access [4143]. These differences reflect a gap in the implementation of the “E” component of the SAFE strategy, which needs on-going emphasis in this region.

Trachoma and Poverty

We have found clear evidence from each measure that even within trachoma-endemic communities individuals and households affected by trichiasis are significantly economically poorer than those that are not. Within endemic communities some individuals or families appear to be more severely affected by the disease and develop sight-threatening complications. This raises the important question of whether the association between poverty and trichiasis arises from a general state of impoverishment or whether there are a number of critical factors that primarily drive the relationship that might be amenable to focused intervention. The data we present here suggest that the relationship between poverty and trachoma could possibly be bidirectional.

Poverty may contribute to trachoma. This study provides evidence that even within superficially homogeneous endemic communities relative poverty plays a major part in the vulnerability of families to scarring disease. Firstly, trichiasis cases were more likely than the controls to come from households where the main family job is daily labouring and from families with no or lower formal education. Both of these factors have a major influence on income and health awareness, which in turn increase the vulnerability of the family to trachoma. Consistent with this, studies from Malawi, Tanzania and Ethiopia identified that children from lower socio-economic households had a higher prevalence of active trachoma than their counterparts indicating an association between poverty and active trachoma [10,44,45]. Secondly, previously described risk factor associations for active trachoma such as crowding and poor access to latrine, characterised the households of the trichiasis cases in this study. Such conditions are believed to promote the transmission of Chlamydia trachomatis within endemic communities, sustaining higher prevalence levels. Poorer households and communities may be less likely to have either the resources or the awareness to access treatment and sustain a sufficiently hygienic environment to control trachoma [8,17,46,47]. Households with higher income were more likely to have a latrine than their counterparts in a study conducted in the same area [48].

Trachoma may also contribute to poverty. Poor health frequently results in loss of productivity through disability and diversion of resources [11]. Trichiasis and its associated visual impairment probably lead to a loss of income, exacerbating pre-existing poverty in a “vicious cycle” [12,13]. Previously healthy and productive adults can be rendered dependent on others, unable to work or fully care for themselves due to pain, photophobia or visual impairment [13]. We found clear evidence of reduced activity and participation among trichiasis cases. Trichiasis cases were less likely than the controls to participate in productive household activities, outdoor activities (shopping/marketing, fetching wood and water) and agricultural activities (farming, animal husbandry and processing agricultural products). The stratified analysis found trichiasis cases with normal vision are less likely to participate in outdoor and agricultural activities than controls. This is consistent with a study of Tanzanian women with trichiasis without visual impairment, who had a degree of functional limitation which was comparable to those with visual impairment [14]. We found evidence that households with fewer economically productive adults and where the family head had trichiasis tended to be poorer. Conversely, households where trichiasis cases participated in agricultural activities were better off. Even where the trichiasis cases were undertaking specific activities, they reported much more difficulty and greater need for assistance than the controls. Similarly in another study, trichiasis cases reported difficulty in performing day-to-day farming activities [49]. These observations all point towards households with someone with trichiasis being under greater financial strains through reduced income contribution and greater needs and dependence of the person with trichiasis. The burden of disability caused by trachoma has been estimated between 171,000 and 1.3 million DALYs, with economic losses of 5–8 billion USD/year [4,12,13]. The economic loss from trichiasis alone due to lost productivity was estimated to be 3 billion USD/year [12,13].

Study Strengths

This study comprehensively assesses the relationship between trachoma and economic poverty using four different measures, with a robust process to select suitable community controls. The asset index quantifies the long-term economic welfare of trachoma affected communities, which is important as trachoma and its sequelae are probably related to long-term SES [19,20]. The asset index has the practical advantage that it is much less affected by recall or measurement bias during data collection [19]. Most of the housing characteristics, utilities and durable assets were collected through direct observation minimising miss-measurement. Broad ranges of asset data were collected increasing the power of the study in the following ways. Clumping and truncation, potential problems that can arise with PCA of asset data and compromise its suitability for defining socio-economic strata, did not occur when all asset indices were combined into a single index. This indicates that the data from this study is sufficient to measure economic status and effectively infer inequality between different socio-economic strata and that in this region assessment of economic status by asset measurement requires a wider pool of parameters, particularly including agricultural assets. Encouragingly, the asset based poverty measure was moderately and strongly correlated with the self-rated and peer-rated wealth measures.

Limitations of the Study

Poverty is a complex multidimensional problem with many causes and manifestations. Therefore there are many ways in which poverty can be measured. Here we only examined the economic aspect using relative measures such as low asset ownership. We use the first principal component (PC1) to measure socio-economic status. However, there is no clear description of the number of principal components to use and often the factor scores derived from the other principal components are difficult to interpret [27]. Despite the comparability of the amount of variance explained by PC1 with other studies, there is uncertainty whether the first component alone sufficiently explains all the pertinent variation. Asset scores are usually developed to be locally relevant, to allow ranking of people within the same community with respect to poverty. Unfortunately, socio-economic classifications based on asset ownership quintiles measure relative poverty within a given context and face the limitation of lacking international comparability. Therefore, between region or country comparison of SES should be done with caution [28]. We did not collect consumption or expenditure data, and so were not able to assess absolute poverty levels.

Although a community based screening method was used to identify trichiasis cases, it is possible that some cases might have been missed, which could potentially introduce non-response bias. Similarly, it is possible that some potential controls were not listed by the sub-village administrators. Self and peer-rated wealth are subjective measures, which might have suffered from the tendency to favour ranking households in the middle of the distribution. The activity participation data relied on the participant’s recall ability on what s/he had done in the last week. Finally, our results suggest that a bidirectional relationship may possibly exist between trachoma and poverty. However, the authors recognise that inference about causality is speculative as it is not possible to draw firm conclusions from a cross-sectional observational study such as this.


In this study we found a clear association between trichiasis and household economic poverty by all three economic measures. Trichiasis cases were more likely to have economically poor households and less likely to participate in productive activities regardless of visual impairment, more likely to report difficulty in performing productive activities and more likely to need assistance in performing activities than controls. These suggest that the causative relationships between poverty and trachoma may possibly involve bidirectional interaction: poor households are more affected by trachoma and the scarring sequelae of trachoma and trichiasis reduces productivity even prior to the development of visual impairment, which might lead to additional poverty.

These data are anticipated to be useful in advocacy and to support programme leaders and funders to secure resources to promote trachoma prevention linked to socio-economic development in trachoma-endemic communities. Implementation of the full SAFE strategy in the context of general improvements might lead to a virtuous cycle of improving health and wealth. Trachoma is a good proxy of inequality within communities and it could be used to target and evaluate interventions for health and poverty alleviation. Measuring the effect of trichiasis surgery on household economic poverty through longitudinal studies would provide an indication of the relative contribution of trichiasis to poverty, as improved health potentially leads to improved productivity and income.

Supporting Information

S1 Checklist. STROBE Checklist.



S1 Table. Summary statistics and principal component factor scores for asset variables used in the Principal Component Analysis (PCA).




We are most grateful to the trachoma control program in Amhara National Regional State of Ethiopia, which is collaboration between the Regional Health Bureau and the Lions-Carter Center SightFirst Initiative; the research team in Bahirdar, Amhara Region which did the entire field work of the study; the study participants; West Gojam Zone Health Office and the respective Woreda Health Offices, and particularly Mecha Woreda Administration Office.

Author Contributions

Conceived and designed the experiments: EH PME HK SP HW MJB. Performed the experiments: EH TW SA ZT MZ ZZ MJB. Analyzed the data: EH HAW MJB. Wrote the paper: EH TW SA ZT MZ ZZ KC PME HK RLB DCWM SNR SP HAW MJB.


  1. 1. Pascolini D, Mariotti SP (2012) Global estimates of visual impairment: 2010. British Journal of Ophthalmology 96: 614–618. doi: 10.1136/bjophthalmol-2011-300539. pmid:22133988
  2. 2. World Health Organization (2014) WHO Alliance for the Global Elimination of Blinding Trachoma by the year 2020. Progress report on elimination of trachoma, 2013. Wkly Epidemiol Rec 89: 421–428. pmid:25275153
  3. 3. World Health Organization (2012) Global WHO alliance for the elimination of blinding trachoma by 2020. Weekly Epidemiological Record 87: 161–168. pmid:22574352
  4. 4. Murray CJ, Barber RM, Foreman KJ, Ozgoren AA, Abd-Allah F, et al. (2015) Global, regional, and national disability-adjusted life years (DALYs) for 306 diseases and injuries and healthy life expectancy (HALE) for 188 countries, 1990–2013: quantifying the epidemiological transition. The Lancet. doi: 10.1016/S0140-6736(15)61340-X. pmid:26321261
  5. 5. World Health Organization (2006) Trachoma control: a guide for programme managers.
  6. 6. Taylor HR, Burton MJ, Haddad D, West S, Wright H (2014) Trachoma. The Lancet. doi: 10.1016/s0140-6736(13)62182-0
  7. 7. Kuper H, Solomon AW, Buchan J, Zondervan M, Foster A, et al. (2003) A critical review of the SAFE strategy for the prevention of blinding trachoma. The Lancet Infectious Diseases 3: 372–381. pmid:12781509 doi: 10.1016/s1473-3099(03)00659-5
  8. 8. Wright HR, Turner A, Taylor HR (2007) Trachoma and poverty: unnecessary blindness further disadvantages the poorest people in the poorest countries. Clin Exp Optom 90: 422–428. pmid:17958564 doi: 10.1111/j.1444-0938.2007.00218.x
  9. 9. Emerson PM, Cairncross S, Bailey RL, Mabey DCW (2000) Review of the evidence base for the ‘F’ and ‘E’ components of the SAFE strategy for trachoma control. Tropical Medicine & International Health 5: 515–527. doi: 10.1046/j.1365-3156.2000.00603.x
  10. 10. Jansen E, Baltussen RM, van Doorslaer E, Ngirwamungu E, Nguyen MP, et al. (2007) An Eye for Inequality: How Trachoma Relates to Poverty in Tanzania and Vietnam. Ophthalmic Epidemiology 14: 278–287. pmid:17994437 doi: 10.1080/09286580701299403
  11. 11. Naidoo K (2007) Poverty and blindness in Africa. Clinical and Experimental Optometry 90: 415–421. pmid:17958563 doi: 10.1111/j.1444-0938.2007.00197.x
  12. 12. Frick KD, Hanson CL, Jacobson GA (2003) Global burden of trachoma and economics of the disease. Am J Trop Med Hyg 69: 1–10.
  13. 13. Burton MJ, Mabey DCW (2009) The Global Burden of Trachoma: A Review. PLoS Negl Trop Dis 3: e460. doi: 10.1371/journal.pntd.0000460. pmid:19859534
  14. 14. Frick KD, Melia BM, Buhrmann RR, West SK (2001) Trichiasis and disability in a trachoma-endemic area of Tanzania. Archives of ophthalmology 119: 1839. pmid:11735797 doi: 10.1001/archopht.119.12.1839
  15. 15. Dandona L, Dandona R, Srinivas M, Giridhar P, Vilas K, et al. (2001) Blindness in the Indian state of Andhra Pradesh. Invest Ophthalmol Vis Sci 42: 908–916. pmid:11274066
  16. 16. Kuper H, Polack S, Eusebio C, Mathenge W, Wadud Z, et al. (2008) A Case-Control Study to Assess the Relationship between Poverty and Visual Impairment from Cataract in Kenya, the Philippines, and Bangladesh. PLoS Med 5: e244. doi: 10.1371/journal.pmed.0050244. pmid:19090614
  17. 17. Gilbert CE, Shah SP, Jadoon MZ, Bourne R, Dineen B, et al. (2008) Poverty and blindness in Pakistan: results from the Pakistan national blindness and visual impairment survey. BMJ 336: 29–32. pmid:18087076 doi: 10.1136/
  18. 18. Alavi Y, Kuper H (2010) Evaluating the Impact of Rehabilitation in the Lives of People with Disabilities and their Families in Low and Middle Income Countries: A Review of Tools. London School of Hygiene & Tropical Medicine, UK.
  19. 19. McKenzie DJ (2005) Measuring inequality with asset indicators. Journal of Population Economics 18: 229–260. doi: 10.1007/s00148-005-0224-7
  20. 20. Filmer D, Pritchett LH (2001) Estimating wealth effects without expenditure data—or tears: an application to educational enrollments in states of India. Demography 38: 115–132. pmid:11227840 doi: 10.1353/dem.2001.0003
  21. 21. Emerson PM, Ngondi J, Biru E, Graves PM, Ejigsemahu Y, et al. (2008) Integrating an NTD with one of “the big three”: combined malaria and trachoma survey in Amhara Region of Ethiopia. PLoS neglected tropical diseases 2: e197. doi: 10.1371/journal.pntd.0000197. pmid:18350115
  22. 22. Grosh ME, Glewwe P, Reconstruction IBf, Development. (2000) Designing household survey questionnaires for developing countries: lessons from 15 years of the Living Standards Measurement Study: World Bank Washington, DC.
  23. 23. Polack S, Kuper H, Eusebio C, Mathenge W, Wadud Z, et al. (2008) The Impact of Cataract on Time-use: Results from a Population Based Case-Control Study in Kenya, the Philippines and Bangladesh. Ophthalmic Epidemiology 15: 372–382. doi: 10.1080/09286580802478716. pmid:19065430
  24. 24. Bastawrous A, Rono HK, Livingstone IT, et al. (2015) Development and validation of a smartphone-based visual acuity test (peek acuity) for clinical practice and community-based fieldwork. JAMA Ophthalmology. doi: 10.1001/jamaophthalmol.2015.1468
  25. 25. Dawson CR, Jones BR, Tarizzo ML (1981) Guide to trachoma control in programmes for the prevention of blindness: Geneva, Switzerland.
  26. 26. Prakongsai P (2006) An application of the asset index for measuring household living standards in Thailand. International Health Policy Program (IHPP)-Thailand.
  27. 27. Vyas S, Kumaranayake L (2006) Constructing socio-economic status indices: how to use principal components analysis. Health Policy and Planning 21: 459–468. pmid:17030551 doi: 10.1093/heapol/czl029
  28. 28. Gwatkin DR, Rutstein S, Johnson K, Suliman E, Wagstaff A, et al. (2000) Socio-economic differences in health, nutrition, and population. Ethiopia: 2000.
  29. 29. Benjamini Y, Hochberg Y (1995) Controlling the false discovery rate: a practical and powerful approach to multiple testing. Journal of the Royal Statistical Society Series B (Methodological): 289–300.
  30. 30. Polack S, Eusebio C, Fletcher A, Foster A, Kuper H (2010) Visual Impairment from Cataract and Health Related Quality of Life: Results from a Case-Control Study in the Philippines. Ophthalmic Epidemiology 17: 152–159. doi: 10.3109/09286581003731536. pmid:20455844
  31. 31. Rajak SN, Habtamu E, Weiss HA, Kello AB, Gebre T, et al. (2011) Absorbable versus silk sutures for surgical treatment of trachomatous trichiasis in Ethiopia: a randomised controlled trial. PLoS medicine 8: e1001137. doi: 10.1371/journal.pmed.1001137. pmid:22180732
  32. 32. World Bank (2001) World development report 2000/2001: attacking poverty. New York, Oxford University Press), 19: 311.
  33. 33. Haughton JH, Khandker SR (2009) Handbook on poverty and inequality: World Bank Publications.
  34. 34. World Bank (2015) A Measured Approach to Ending Poverty and Boosting Shared Prosperity: Concepts, Data, and the Twin Goals. Washington DC.
  35. 35. World Bank (January 2015) Ethiopia Poverty Assessment 2014.
  36. 36. Houweling TA, Kunst AE, Mackenbach JP (2003) Measuring health inequality among children in developing countries: does the choice of the indicator of economic status matter? International journal for equity in health 2: 8. pmid:14609435
  37. 37. Howe LD, Hargreaves JR, Gabrysch S, Huttly SRA (2009) Is the wealth index a proxy for consumption expenditure? A systematic review. Journal of Epidemiology and Community Health 63: 871–877. doi: 10.1136/jech.2009.088021. pmid:19406742
  38. 38. Rajak SN, Habtamu E, Weiss HA, Kello AB, Gebre T, et al. (2011) Surgery Versus Epilation for the Treatment of Minor Trichiasis in Ethiopia: A Randomised Controlled Noninferiority Trial. PLoS Med 8: e1001136. doi: 10.1371/journal.pmed.1001136. pmid:22180731
  39. 39. Merbs SL, West SK, West ES (2005) Pattern of Recurrence of Trachomatous Trichiasis After Surgery: Surgical technique as an explanation. Ophthalmology 112: 705–709. pmid:15808265 doi: 10.1016/j.ophtha.2004.10.037
  40. 40. Bowman RJ, Faal H, Jatta B, Myatt M, Foster A, et al. (2002) Longitudinal study of trachomatous trichiasis in The Gambia: barriers to acceptance of surgery. Invest Ophthalmol Vis Sci 43: 936–940. pmid:11923231
  41. 41. Burton MJ, Holland MJ, Jeffries D, Mabey DC, Bailey RL (2006) Conjunctival chlamydial 16S ribosomal RNA expression in trachoma: is chlamydial metabolic activity required for disease to develop? Clin Infect Dis 42: 463–470. pmid:16421789 doi: 10.1086/499814
  42. 42. Emerson PM, Lindsay SW, Alexander N, Bah M, Dibba S- M, et al. Role of flies and provision of latrines in trachoma control: cluster-randomised controlled trial. The Lancet 363: 1093–1098. doi: 10.1016/s0140-6736(04)15891-1
  43. 43. Burton MJ, Holland MJ, Makalo P, Aryee EAN, Alexander NDE, et al. Re-emergence of Chlamydia trachomatis infection after mass antibiotic treatment of a trachoma-endemic Gambian community: a longitudinal study. The Lancet 365: 1321–1328. doi: 10.1016/s0140-6736(05)61029-x
  44. 44. Tielsch JM, West KP, Katz J, Keyvan-Larijani E, Tizazu T, et al. (1988) The Epidemiology of Trachoma in Southern Malawi. The American Journal of Tropical Medicine and Hygiene 38: 393–399. pmid:3354773
  45. 45. Ketema K, Tiruneh M, Woldeyohannes D, Muluye D (2012) Active trachoma and associated risk factors among children in Baso Liben District of East Gojjam, Ethiopia. BMC public health 12: 1105. doi: 10.1186/1471-2458-12-1105. pmid:23259854
  46. 46. Peters DH, Garg A, Bloom G, Walker DG, Brieger WR, et al. (2008) Poverty and access to health care in developing countries. Annals of the New York Academy of Sciences 1136: 161–171. pmid:17954679 doi: 10.1196/annals.1425.011
  47. 47. Wagstaff A (2002) Poverty and health sector inequalities. Bulletin of the world health organization 80: 97–105. pmid:11953787
  48. 48. Awoke W, Muche S (2013) A cross sectional study: latrine coverage and associated factors among rural communities in the District of Bahir Dar Zuria, Ethiopia. BMC public health 13: 99. doi: 10.1186/1471-2458-13-99. pmid:23374236
  49. 49. Wolle MA, Cassard SD, Gower EW, Munoz BE, Wang J, et al. (2011) Impact of Trichiasis surgery on physical functioning in Ethiopian patients: STAR trial. Am J Ophthalmol 151: 850–857. doi: 10.1016/j.ajo.2010.10.039. pmid:21333268