Impact of Trichiasis Surgery on Physical Functioning in Ethiopian Patients: STAR Trial


To evaluate the physical functioning of Ethiopian trichiasis surgery patients before and 6 months after surgery.


Nested cohort study.


This study was nested within the Surgery for Trichiasis, Antibiotics to Prevent Recurrence (STAR) clinical trial conducted in Ethiopia. Demographic information, ocular examinations, and physical functioning assessments were collected before and 6 months after surgery. A single score for patients’ physical functioning was constructed using Rasch analysis. A multivariate linear regression model was used to determine if change in physical functioning was associated with change in visual acuity.


Of the 438 participants, 411 (93.8%) had both baseline and follow-up questionnaires. Physical functioning scores at baseline ranged from −6.32 (great difficulty) to +6.01 (no difficulty). The percentage of participants reporting no difficulty in physical functioning increased by 32.6%; the proportion of participants in the mild/no visual impairment category increased by 8.6%. A multivariate linear regression model showed that for every line of vision gained, physical functioning improves significantly (0.09 units; 95% CI: 0.02–0.16).


Surgery to correct trichiasis appears to improve patients’ physical functioning as measured at 6 months. More effort in promoting trichiasis surgery is essential, not only to prevent corneal blindness, but also to enable improved functioning in daily life.

Trachoma, an ocular infection caused by Chlamydia trachomatis, is the leading infectious cause of blindness worldwide. Approximately 40.6 million children have active trachoma; 8.2 million, mainly adults, are suffering from the late sequela, trichiasis, and about 1.3 million adults are blinded from the end stage, corneal opacity.

Trachoma is seen in rural, economically underdeveloped areas where living conditions are crowded and good water supplies and basic sanitation services are lacking. Active disease is seen primarily in young children, and repeated bouts lead to conjunctival scarring in young adults. In adults, the prevalence and severity of scarring increases; extensive scarring may result in entropion and trichiasis. Trichiasis requires lid surgery to correct in-turned eyelashes which, if uncorrected, places these individuals at high risk of visual loss and blindness from corneal opacification.

Ethiopia has one of the highest prevalences of trachoma in the world. A national survey on blindness, low vision, and trachoma conducted in 2005 showed active trachoma prevalence as high as 60% among children aged less than 10 years in certain hyperendemic regions. The prevalence of trichiasis in these regions was as high as 5% among adults. Estimates suggest that greater than 1 million individuals in Ethiopia are living with trichiasis.

The World Health Organization (WHO) has adopted the SAFE (Surgery for trichiasis, Antibiotic use, Face washing, and Environmental change) strategy with the goal of the global elimination of blinding trachoma by the year 2020. A number of different surgical techniques have been used to correct trichiasis; however, the WHO recommends bilamellar tarsal rotation, as this technique has been shown in clinical trials to be effective.

Surgery is performed to prevent further visual loss, but recent research has shown that surgery also improves photophobia and ocular pain, and appears to improve visual acuity. Frick and associates showed that in women, trichiasis was associated with excess functional limitation even without visual loss. When vision loss was present in combination with trichiasis, functional limitation was greater than has been reported with cataracts. However, whether trichiasis surgery translates into improvement in daily life has not been investigated. We hypothesize that trichiasis surgery results in an improvement in self-reported physical functioning associated with improvement in visual acuity in a cohort of trichiasis patients who received surgery in rural Ethiopia.


This study was nested within the Surgery for Trichiasis, Antibiotics to Prevent Recurrence (STAR) clinical trial. The STAR trial is a randomized, single-masked clinical trial of antibiotic use in 1452 patients who received trichiasis surgery in the Wolayta Soddo zone of the Southern Nations, Nationalities, and Peoples Region (SNNPR) of Ethiopia. The primary aim was to determine whether postoperative treatment with azithromycin reduces trichiasis recurrence as compared to topical tetracycline within 1 year after surgery.

The methods have been described in detail elsewhere. In summary, patients aged 18 years or older with trichiasis present in at least 1 eye, with no previous report of trichiasis surgery in that eye, were eligible for enrollment into the study. All patients had bilamellar tarsal rotation surgery performed by trained integrated eye care workers, all of whom were certified using WHO guidelines. Participants were randomly assigned to postoperative azithromycin or tetracycline treatment groups; the primary trial results have been reported previously.

Examinations were performed at each visit to assess presence of trichiasis and to evaluate trichiasis severity. The 438 participants enrolled during the third recruitment period (October-December 2003) were interviewed using a physical functioning questionnaire, which included perceived difficulty with activities, before and 6 months after surgery. These 438 participants comprise the cohort used for this sub-study.

Visual acuity was measured at baseline and 6 months post surgery. Monocular acuity was measured for each eye, with the participant’s presenting refractive correction, using a tumbling “E” chart patterned on a standard Early Treatment for Diabetic Retinopathy Study chart. Trichiasis was defined according to WHO criteria as the presence of 1 or more lashes touching the globe and/or evidence of epilation. Entropion was graded as mild (all lash bases visible), moderate (some lash bases visible), or severe (no lash bases visible).

Physical functioning was assessed by the administration of a questionnaire. A trained interviewer interviewed the study participants in either Amharic or Wolaytinga, the national and local language, respectively. The questionnaire included activities identified as important for daily functioning in rural Ethiopia. The questionnaire was based on a previous quality-of-life instrument used in rural Tanzania with Ethiopia-specific factors added, based on field testing. The items included walking around the village, observing obstacles while walking, recognizing faces, gathering wood, chopping wood, milking cows, helping with the farm, making injera (local bread), grinding coffee, and caring for children. The latter 3 tasks were specific to women. The possible answers for the first 3 questions ranged along a 5-point scale: no difficulty, a little difficulty, some difficulty, a lot of difficulty, unable to do the activity. The possible answers for the remaining 7 questions included 2 additional answer choices: never did the activity, and stopped doing the activity for reasons other than vision or trichiasis.

Initial Rasch analysis of the questionnaire responses indicated that the categories “a little difficulty” and “some difficulty” were not discernibly different. Thus, prior to any further analyses, the physical functioning questionnaire responses of “a little” and “some” were collapsed into 1 category, turning the 5-point scale into a 4-point scale.

The distribution of responses for all items was examined. Any item for which any degree of difficulty was indicated (none, some, a lot) at baseline but then the response “never did” was given at follow-up was re-coded to “never did” at both time points (n = 72 responses). Any item for which the response “cannot do” was given at baseline and “never did” was given at follow-up was re-coded to “cannot do” at both time points (n = 16 responses). Any item for which “stopped doing for other reasons” was given at baseline and “never did” was given at follow-up was re-coded to “stopped doing for other reasons” (n = 12 responses). These discrepancies indicate the presence of a possible misunderstanding of the meaning of the “never did” response, namely the clear difference between “never did the activity” (meaning never, even before trichiasis) versus “can no longer perform the activity” (meaning never did since trichiasis or eye problem started). Gender-specific evaluations of questionnaire responses were conducted since men and women partake in different activities. The questions involving chopping wood for women and milking cows for men were removed from all further analyses since insufficient numbers of persons in the respective genders undertook these activities.

A single score for patients’ physical functioning was constructed using Andrich’s rating scale model, one type of Rasch model. The Rasch model allows responses from different questions (called items) representing different severity or weights to be summed. Weights do not have to be assigned to items in advance. The probability of how a person responds to a rating category of a particular item depends on the characteristic of the item (ie, difficulty of a given question) and the person (ie, perceived impairment in physical functioning from a given disability), plus the response threshold (ie, expected response generated by the Rasch model). The estimated scores were generated from the Rasch analysis on a common logit (log-odds) scale.

The resulting Rasch model was evaluated to ensure that the assumptions were satisfied. Two of the 3 female-specific items (caring for children and making injera) did not fit into the single scale and were excluded. Subgroup analysis based on gender was carried out and showed consistency among the subgroups; this consistency allowed us to combine the subgroups for subsequent analyses. The remaining 7 items applicable to women and the 6 items for men showed acceptable fit and were kept in the final Rasch model to generate the final physical functioning scores. WINSTEPS (MESA Press, Chicago, Illinois, USA) was used to carry out the Rasch analyses.

Further data analyses were performed using STATA version 11.0 (STATA-Corp LP, College Station, Texas, USA). Visual acuity was scored as the number of letters read and then converted to logMAR values. LogMAR values were converted to Snellen values for visual impairment categorization. Visual impairment was categorized as: mild or no visual impairment, visual acuity better than or equal to 20/70; moderate/severe visual impairment, visual acuity worse than 20/70 but equal to or better than 20/400; or blindness, visual acuity worse than 20/400. Bivariate relationships were explored to examine differences in baseline characteristics by those who had data at both baseline and follow-up versus those who were missing data at 1 of these time points, and by those who reported no difficulty with physical functioning at baseline versus those who reported some difficulty. We compared differences between the groups using the Fisher exact test or a test for trend, if the variable was ordinal.

Changes in visual acuity between baseline and follow-up were assessed. The difference in baseline characteristics and change in vision by change in physical functioning was examined using 1-way analysis of variance for continuous variables, test for trend for ordinal variables, and Fisher exact test for proportions. Change in vision was defined in 2 different ways: first, as the difference in acuity between the better eye at baseline and the better eye at follow-up (allowing the eye to change); second, as the maximum difference between baseline and follow-up of either the right or left eye. For our final analyses, we chose the second definition, as it appeared to reflect the effect of surgery on vision more completely.

A multivariate linear regression model was used to determine if change in physical functioning was associated with change in visual acuity, after controlling for age, gender, severity of entropion, vision at baseline, and baseline physical functioning.


Of the 438 participants, 18 (4.1%) did not have baseline questionnaires, and 9 (2.1%) were not present at follow-up; 5 of 9 had developed recurrent trichiasis before the 6-month follow-up. Thus, 411 participants had both baseline and follow-up questionnaires. Most of these 411 participants were female (74.5%). At baseline, 131 of the 411 participants (31.9%) had severe entropion, and 330 (80.3%) had surgery on both eyelids ( Table 1 ). When comparing these 411 participants to the 27 participants who were missing a questionnaire at either baseline or follow-up, only mean age was different. The mean age of participants who were missing at least 1 questionnaire was older than those who answered both questionnaires (55 vs 48 years; P = .01).


Baseline Characteristics of Trichiasis Surgery Patients According to Whether Functional Status Was Measured at Both Baseline and the 6-Month Follow-up or Was Missing at Either Time Point

Functional Status Measured at Both Visits Functional Status Missing at 1 Visit P Value
Total participants 93.8% (411) 6.2% (27)
Age (years), mean (SD) 48 (14) 55 (13) .01
Gender, % (N) a .65
Male 25.6% (105) 29.6% (8)
Female 74.5% (306) 70.4% (19)
Entropion, % (N) a .07 b , c
Mild 52.6% (216) 70.4% (19)
Moderate 15.6% (64) 11.1% (3)
Severe 31.9% (131) 18.5% (5)
Trichiasis recurrence at 6 months, % (N) 1.5% (6) N/A N/A
Trichiasis recurrence before 6 months % (N) N/A 18.5% (5) N/A
Bilateral surgery, % (N) 80.3% (330) 74.1% (20) .46
Mean (SD) baseline physical functioning score 3.00 (3.02) 3.03 (2.62) d .98
Level of visual impairment, e % (N) .73 c
Mild/no visual impairment 62.2% (253) 66.7% (6)
Moderate/severe visual impairment 26.8% (109) 11.1% (1)
Blind 11.1% (45) 22.2% (2)

a Totals may not equal 100% due to rounding.

b The moderate and severe entropion groups were collapsed prior to significance testing.

c Test for trend.

d Consists of 9 people; 18 were missing baseline data.

e Four people missing baseline visual acuity data.

The final physical functioning scores ranged from −6.32 (great difficulty) to +6.01 (no difficulty). The mean physical functioning score at baseline was 3.00. The distribution of the baseline physical functioning scores is shown in Figure 1 . The distribution was skewed because greater than one-third of the study participants reported no difficulty with physical functioning at baseline. Baseline characteristics comparing those with no difficulty to those with some difficulty are presented in Table 2 . Age, gender composition, and the percentage that underwent bilateral surgery were similar in the 2 groups; degree of entropion and visual acuity differed significantly. Participants with no difficulty at baseline were more likely to have mild entropion and mild or no visual impairment compared to those with some difficulty at baseline.


Distribution of Rasch scores for physical functioning of trichiasis patients at baseline.


Baseline Comparison of Trichiasis Surgery Patients who Reported No Difficulty on Any Item of Physical Functioning to Those who Reported Any Difficulty

No Difficulty Any Difficulty P Value
Total participants 38.9% (160) 61.1% (251)
Age (years), mean (SD) 47 (13) 49 (14) .10
Gender, % (N) .36
Male 28.1% (45) 23.9% (60)
Female 71.9% (115) 76.1% (191)
Entropion, % (N) .01 a
Mild 60.0% (96) 47.8% (120)
Moderate 14.4% (23) 16.3% (41)
Severe 25.6% (41) 35.9% (90)
Trichiasis recurrence by 6 months, % (N) 0.6% (1) 2.0% (5) .41
Bilateral surgery, % (N) 76.3% (122) 82.9% (208) .13
Level of visual impairment, b , c % (N) <.0001 a
Mild/no visual impairment 74.8% (119) 54.0% (134)
Moderate/severe visual impairment 22.0% (35) 29.8% (74)
Blind 3.1% (5) 16.2% (40)

a Test for trend.

b Totals may not equal100% due to rounding.

c Four people missing baseline vision data.

The mean improvement in the physical functioning score from baseline to 6 months was 2.08 units ( Table 3 ). The percentage of participants reporting no difficulty in physical functioning increased by 32.6%. The proportion of participants in the mild/no visual impairment category also increased, commensurate with the decreased proportion of people in the moderate/severe visual impairment and blind categories.


Vision and Physical Functioning Score of Trichiasis Surgery Patients at Baseline and 6 Months Following Surgery

Baseline Level of Vision and Functioning 6 Months After Surgery: Level of Vision and Functioning Change in Vision and Functioning
Level of visual impairment, a % (N)
Mild/no visual impairment 62.2% (253) 70.8% (291) 8.6%
Moderate/severe visual impairment 26.8% (109) 23.6% (84) −3.2%
Blind 11.1% (45) 5.6% (23) −5.5%
Score of physical functioning
No difficulty, % (N) 38.9% (160) 71.5% (294) 32.6%
Mean (SD) 3.00 (3.02) 5.09 (1.64) 2.08 (2.63)
Median 3.70 5.97 1.41

a Four people missing baseline vision data; totals may not equal 100% due to rounding.

The change in mean physical functioning score from baseline to follow-up by level of visual impairment is presented in Table 4 . Across all categories of visual impairment, physical functioning improved; the blind group had the biggest gain in physical function (3.30 ± 0.52 units).

Jan 16, 2017 | Posted by in OPHTHALMOLOGY | Comments Off on Impact of Trichiasis Surgery on Physical Functioning in Ethiopian Patients: STAR Trial

Full access? Get Clinical Tree

Get Clinical Tree app for offline access