SEARCH

SEARCH BY CITATION

Keywords:

  • trichiasis;
  • trachoma;
  • surgery;
  • community;
  • village;
  • compliance

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Summaryintroduction  Surgery for trachomatous trichiasis prevents blindness and is advocated by the WHO as part of the SAFE strategy for the global elimination of trachoma. We conducted a randomised community trial to investigate the effect of providing surgery in villages on surgical uptake in The Gambia.

methods  56 villages from two divisions were assigned to eight pairs of clusters matched by geographical division and proximity. One cluster from each pair was randomly assigned to receive village-based surgery and the other cluster health centre-based surgery. Outcome measures were uptake rates and surgical results after 1 week and 3 months. The paired t-test was used to analyse the results.

results  Overall uptake was 66% in the village-based clusters and 44% in the health centre-based clusters. Subjects in the village-based surgery arm had significantly shorter journey times (= 0.01) and lower costs (= 0.002). The mean difference in absolute acceptance rates of surgery was 20% better in village-based clusters (95% CI –9 to + 49%, = 0.15), which would equate to an improvement of 45% (95% CI −20% to 120%) on the average acceptance rates of 44% in the health centre-based group.

conclusion  These results strongly suggest better surgical uptake when surgery is provided in patients' villages due to lower cost to the patient, time saved and less fear of the operation.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Trachoma is a chronic conjunctivitis caused by Chlamydia trachomatis which may have sight-threatening complications such as trichiasis and corneal scarring. It remains the world's commonest form of both infectious and preventable blindness, mainly affecting developing countries. The World Health Organization has estimated that 146 million children and adults have active inflammatory trachoma, with over 500 million exposed to the risk of disease, and that an additional 10 million people have trichiasis and urgently need surgery ( WHO 1997).

The World Health Organization has targeted the Global Elimination of Trachoma as a blinding public health problem by the year 2020 (GET 2020) and has endorsed the SAFE strategy: Surgery for entropion and trichiasis, Antibiotic treatment for active infection, and the promotion of both Facial cleanliness and Environmental improvement to reduce transmission in order to achieve this goal ( WHO 1998). Surgery is the one component of the SAFE strategy which has been shown to prevent blindness ( Reacher et al. 1992 ) and usually results in immediate and dramatic relief of discomfort. For these reasons, surgery is usually the first component of the SAFE strategy to be introduced into a community, and is often important in gaining support for the other elements of SAFE which may not be perceived to have such an immediate and obvious benefit. However, acceptance of surgery in some affected communities has been low, reaching only 18% in Tanzania and 35% in Malawi ( Courtright 1994; West et al. 1994 ). The need to investigate and reduce barriers to trichiasis surgery was identified as a research priority by the WHO Alliance ( WHO 1996).

For the past 10 years, the Gambian National Eye Care Programme has provided trichiasis surgery in two major health centres and two hospitals distributed throughout the country on a weekly basis. Outreach surgery is also performed from each of these four centres in at least two rural health centres on a monthly basis. The standard operation fee is 25 dalasis (approximately $2.5), which is a fixed government levy paid by the patient to the hospital or health centre funds rather than the eye care programme itself. Surgery is performed by trained nurses with 3 years basic training and a 1-year ophthalmic nursing course including surgical techniques and follow-up of at least 10 trichiasis operations. In major centres surgery is performed by ophthalmic assistants who, in addition to the above, have had a further 2 years clinical and surgical training.

The standard surgical technique used in The Gambia is the tarsal rotation procedure, which Bog et al. (1993) found to be safe and effective when performed by a trained nurse in the community. It is similar but not identical to the bilamellar tarsal rotation whose effectiveness was demonstrated in a randomised controlled trial ( Reacher et al. 1992 ).

The availability of trichiasis surgery in rural health centres has improved accessibility, with a national average 30–40 minute journey time (by public transport) to the nearest surgical centre. This was corroborated by two nationwide prevalence sampling surveys (Foal et al. 1989; Dolin et al. 1998 ) which have enabled the programme both to estimate the size of the problem and to identify the areas where trichiasis remains a priority. Despite these achievements, a recent survey indicated that there are still 10 000 patients with unoperated trichiasis in The Gambia.

To improve surgical acceptance, the national programme decided to provide surgery by community ophthalmic nurses at the village level. This had the dual effect of reducing geographical distance from patients' houses to the site of surgery and of removing the fixed health centre levy. In Tanzania, surgical uptake was low (an improvement of 12% on the original 18%) despite the surgery being provided in villages ( Oliva et al. 1998 ). Therefore, during the introduction of the village-based surgery, a priority for the Gambian programme was to estimate the effect of this strategy on surgical acceptance rates, to determine whether and by how much they could be improved by offering surgery at village level. In this study we used the phased countrywide introduction of village-based surgery to compare uptake of free trichiasis surgery provided in the village with uptake of free trichiasis surgery in health centres by means of a paired cluster randomised trial.

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Population selection

Districts in areas found to have high trichiasis rates in national surveys and where village-based surgery had not been previously available were considered for inclusion. Two districts from North Bank East division and three from Lower River Division were selected as representative of those meeting the criteria.

Cluster randomization

Selected districts were divided into two or four clusters of approximately equal population size using data from the 1993 national census. Cluster allocation and randomization was done before the numbers of trichiasis patients were known because referral for surgery in the village or in a local health centre had to be done at the time of screening. Clusters were paired by district or (when the district contained four clusters) geographical proximity. One cluster from each pair was randomly allocated to receive village-based surgery and the other, health centre-based surgery.

Screening and recruitment

Screening and recruitment was performed by ophthalmic nurses, who attended a standardized training course including community sensitization for village-based lid surgery. Village health workers, village heads (alkalos) and traditional birth attendants were involved in publicizing the screening and surgery, using identical methods in both groups of communities. All patients found to have major trichiasis (five inturned lashes or more) were eligible for recruitment into the study. Consenting subjects completed a questionnaire and underwent a basic ophthalmic examination.

Patients thought to be unsuitable for village-based surgery due to systemic medical complications were excluded from the study and referred for health centre surgery. In clusters allocated to village-based surgery, patients were then referred for surgery either in their own or in an adjacent village no more than 4 km away. In seven of the eight village-based clusters only one central surgical point was required, all patients being within a 4-km radius. In one village-based cluster two surgical points were required. In clusters allocated to health centre surgery, patients were referred to the nearest health centre where lid surgery was performed. Patients in both groups were informed that surgery was free of charge and given a specific time to attend for surgery within two weeks of the screening date. To reduce the possibility that awareness of the availability of village-based surgery might affect uptake in those allocated to health centre surgery, the health centre clusters were screened and treated before the village-based clusters.

Ethical issues

The study was approved by the Medical Research Council and the Gambian Government Ethical Committee. Informed consent was sought from community leaders after an explanation. Subjects who did not accept surgery during the study were offered the more successful treatment strategy after completion of the trial. Incidental ophthalmic or medical conditions detected during conduct of the trial were treated or referred for appropriate treatment.

Surgery and follow-up

On the day of surgery patient attendance and surgical procedure was recorded on the patients' questionnaire. Surgery was performed by an ophthalmic assistant or by one of five trained ophthalmic nurses. Those participants who underwent surgery had follow-up visits in their villages after one week and three months. They were examined for presence or absence of trichiasis and postoperative complications.

Outcome measures

Three outcome measures were assessed: surgical uptake expressed as the proportion of those known to require surgery who attended for it during the trial period, surgical success as judged by absence of trichiasis at follow up, and postoperative complication rates.

Sample size

Sample size was calculated using the suggestions of Hayes and Bennett (1999) for paired cluster randomised trials based on the paired t-test. It was difficult to predict the size of the study sample because clusters had to be allocated prior to screening for trichiasis, before it was known how many surgical subjects would be found. A concurrently running cohort study in another division (Western) indicated attendance rates of around 10–20% for health centre-based surgery. An improvement in attendance rates from 10 to 30% was felt to be important to the programme and sample sizes were calculated to demonstrate such an increase. Prevalence surveys indicated that each cluster would contain approximately 20 surgical patients. This meant that between four and nine pairs were required, depending on intercluster variability, for 80% power and 95% confidence levels. We opted for eight paired clusters.

Data analysis

The paired t-test was used as described by Hayes and Bennett (1999) for cluster randomised trials to compare attendance rates, mean distance from house to place of surgery and mean return journey cost and time for each cluster pair.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

One hundred and fifty-eight subjects were identified as eligible for the trial. None refused to enter the trial and none were excluded on medical grounds. Numbers of patients recruited to each of the eight cluster pairs together with numbers attending for surgery and attendance rates are shown in Table 1. Mean distances between subjects' houses and place of surgery as well as mean costs and times of return journeys are shown in Table 2. Journey times refer to how long the patient took on foot or by public transport. The mean distance between subjects' houses and the surgical centre was significantly greater (t = 4.3, = 0.004) for patients in the health centre arm than the village-based arm. The mean return journey time was also significantly longer (t = 3.4, = 0.011) in the health centre group than the village-based group. The mean cost of the return journey, too, was significantly greater (t = 5.0, = 0.002) in the health centre group than the village-based group.

Table 1.  Cluster sizes and attendance rates for each arm of the trial Thumbnail image of
Table 2.  Mean distances to surgery and return journey costs and times for each cluster pair Thumbnail image of

Table 3 illustrates baseline comparisons between all patients who were offered village-based vs. health centre-based surgery. The village-based and health centre groups were well balanced with respect to the factors examined in Table 3, except that a greater proportion of the village-based group were of Mandinka ethnicity. Attendance rates for surgery were not significantly different for Mandinka (56%) and non-Mandinka (57%). None of the other factors listed in Table 3 affected attendance for surgery.

Table 3.  Baseline comparisons between patients offered village-based vs. health centre-based surgery Thumbnail image of

The patients referred for village-based surgery (mean age 53 years) were slightly younger than those referred to the health centre (mean age 56 years) but this was not statistically significant. The senior ophthalmic medical assistant and the five community ophthalmic nurses all operated in both environments.

Attendance rates for each cluster pair and overall by intervention are illustrated in Figure 1. Acceptance rates for village-based surgery were better in 6 of the 8 pairs, while one of the remaining pairs had extremely small numbers of subjects. As shown in Fig. 1, the overall acceptance rate was 32/72 (44%) for health centre-based surgery and 57/86 (66%) for village-based surgery, a difference of 22%, with a rate ratio of 1.49 (95% CI 1.11–2.01, χ2 = 6.73, = 0.009) in a crude analysis, alternatively expressed as a 49% improvement in surgical uptake. Analysing according to the paired cluster-randomised design by applying the paired t-test to attendance rates of cluster pairs, the mean difference in absolute acceptance rates of surgery was 20% better for the village-based clusters (95% CI –9 to + 49%, = 0.15), which would equate to an improvement of 45% (95% CI –20% to 120%) on the average acceptance rates of 44% in the health centre-based group.

image

Figure 1. Attendance rates for village vs. health centre based surgery.

Download figure to PowerPoint

Post-operative complications occurred rarely, with 2 cases of lid swelling and one granulation reaction to a suture in the community arm and one case of mild wound sepsis in the health centre arm (no significant difference between the arms). None was present at three months and none was associated with surgical failure. Of the 89 cases operated, 82 had no trichiasis at three months, representing a surgical success rate of 92%. There was no significant difference in success rate between village-based cases (91%) and health centre-based cases (94%).

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Uptake of trichiasis surgery increased when the surgery was provided in the patients' own or neighbouring village rather than in a nearby health centre in 6 of 8 pairs. Although there was an absolute increase of 22% in surgical uptake (representing a 49% improvement) if surgery was provided in the village, analysis respecting the design as a cluster randomised trial suggested a 20% increase in absolute attendance rate, with 95% confidence interval from –9% to 49%; a nondefinitive, but strongly suggestive result.

The main weakness of this study was that the calculated sample size proved to be too small. This was firstly because prevalence of surgical cases in the clusters chosen was lower than anticipated. With an average of 10 trichiasis patients per cluster, 7–14 pairs would have been required for 80% power and 95% confidence. Secondly, attendance in both arms was higher than the anticipated 10–20%, reducing the relative difference between them. This is almost certainly because surgery was provided free to all patients in this trial and reflects the importance of cost as a barrier. Village-based surgery is free but there is a charge of 25 dalasis (approximately $2.5) per eye for minor operations performed at health centres. This fee contributes to the general costs of running the health centre rather than the Eye Care Programme. In practice therefore health centre-based surgery may cost an average patient with bilateral trichiasis $8 ($5 for surgery and $3 for travel for surgery and two follow up visits) compared to nothing for village-based surgery. Per capita GDP in The Gambia is estimated at $320. Because this study was intended to test the effect of surgical location, the surgical fee was not paid by the health centre patients to avoid differences attributable to cost. This study, together with other work in The Gambia, suggests that subsidizing surgery may improve attendance, but further studies to investigate the effect of subsidy vs. cost recovery on acceptance rates are warranted.

The nature of the intervention necessitated a cluster design, but there is no evidence to suggest intracluster correlation of outcome. Individual factors, such as time available and degree of disability could be at least as important in determining attendance for surgery as the behaviour of other patients in the village. The simple rate ratio analysis does not allow for increased correlation within clusters and hence may underestimate the variance, leading to confidence intervals that are too narrow. The paired t-test used above goes to the opposite extreme, treating each cluster as an individual case with no allowance for cluster size. This may conversely overestimate the variance. For instance if pair 7 (the smallest cluster size) is omitted, the absolute mean difference was 26% better in village-based clusters (95% CI –3 to +56%, = 0.07). The confidence interval of the effect of the intervention therefore probably lies between those suggested by individual and cluster analysis.

We do not know why two of the eight pairs showed the reverse trend (though one contained very small numbers) nor why the trend appeared exaggerated in pair 8. Acceptance of surgery may depend on a number of factors including cultural attitudes, pressures of time and work as well as the sensitization and communication skills of individual nurses which inevitably lead to intercluster variation and a reduced power in this type of cluster trial. A matched pair design was chosen to minimize the effect of potential confounding variables such as ethnic group, cultural attitudes to surgery or season (since the intervention was phased in time, district by district).

Despite these limitations we feel the improvement in overall attendance rates and in the rates in village-based clusters in 6 of 8 pairs demonstrates a trend of practical value to decision-makers. None of the measured variables appeared to confound attendance rates in the two arms and offering surgery at village level and the consequent reduction of both logistical and psychological barriers is likely to improve attendance rates. Prior to the introduction of village-based surgery, lid surgery was already offered in a large number of peripheral health centres. All patients in the health centre arm were less than an hour's travel from the health centre, at an average cost of $ 1 for the return journey. Thus although surgery itself was free for both arms of this trial, there was a considerable difference in cost to the patient in favour of village surgery, which may have improved acceptance. Reduced journey times are also likely to contribute to the increased attendance, since ongoing work in The Gambia has revealed that lack of time is a significant barrier to uptake of surgery. In addition, fear may have been reduced by providing surgery in the patients' environment. Hollows (1985) has discussed the need for trachoma control measures to be seen by the community as coming from within it. Surgery is certainly seen as foreign and frightening by many Gambians – one local term for trichiasis surgery means ‘tearing the eye’. Seeing the operation performed on friends and acquaintances in the village may reduce cultural and fear barriers.

A study from Tanzania found low compliance rates in women despite introduction of village-based surgery ( Oliva et al. 1998 ). Although ‘before and after’ studies are more difficult to interpret than randomised controlled trials, many women were unaware that village-based surgery was available. When village-based surgery was introduced in The Gambia, training emphasized the role of community sensitization and publicity, enlisting the support of local village and religious leaders. Even when surgery is being provided on the patient's doorstep, such community support is essential.

Transport is the main source of extra cost to the health care system of providing surgery in the village. However, the equipment necessary for surgery can be carried on a motorcycle, and the costs can be further minimized by integrating surgery with ongoing duties in villages such as screening and health education programmes. If surgical equipment were carried on screening visits and immediate surgery offered to patients identified as needing it, acceptance might be further improved. The cost-efficiency of village-based surgery is likely to decline as the number of patients with trichiasis is reduced by programme activities.

In summary, this study suggests that village-based surgery produces an improvement in attendance for this potentially sight-saving operation, in an area where low compliance is known to be a problem. If applied to The Gambia, with 10 000 cases of unoperated trichiasis, we could estimate that an extra 2000 people would receive this operation by nationwide conversion to village-based surgery. However, our study is nondefinitive and does not completely exclude the possibility that village-based surgery may reduce attendance rates. We recommend the use of village-based surgery with careful community sensitization. We also recommend that other programmes considering the introduction of such a service perform similarly designed randomised controlled trials to measure effectiveness so that meta-analysis may produce more powerful support for the policy. Finally we commend this as an example of programme-initiated operational research in which new service developments are measured for effectiveness not only for the benefit of the national programme involved but for other country programmes who can use such evidence to present to funders to argue for the resources necessary for policies of proven efficacy.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

The authors thank Mr M. Bah and Professor K. McAdam for their co-operation. The work was funded by Sight Savers International.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  • Bog H, Yorston D, Foster A (1993) Results of community based eyelid surgery for trichiasis due to trachoma. British Journal of Ophthalmology 77, 81 83.
  • Courtright P (1994) Acceptance of surgery for trichiasis among rural Malawian women. East African Medical Journal 71, 803 804.
  • Dolin PJ, Faal H, Johnson GJ et al. (1998) Trachoma in The Gambia. British Journal of Ophthalmology 82, 930 933.
  • Faal H, Minassian D, Sowa S, Foster A (1989) National survey of blindness and low vision in the Gambia: Results. British Journal of Ophthalmology 73, 82 87.
  • Hayes RJ & Bennett S (1999) Simple sample size calculation for cluster-randomized trials. International Journal of Epidemiology 28, 319 326.
  • Hollows FC (1985) Community based action for the control of trachoma. Review of Infectious Diseases 7, 777 782.
  • Oliva MS, Munoz B, Lynch M, Mkocha H, West SK (1998) Evaluation of barriers to surgical compliance in the treatment of trachoma. International Ophthalmology 21, 235 241.
  • Reacher MH, Munoz B, Alghassany A, Daar AS, Elbualy M, Taylor HR (1992) A controlled trial of surgery for trachomatous trichiasis of the upper lid. Archives of Ophthalmology 110, 667 674.
  • West S, Lynch M, Munoz B, Katala S, Tobin S, Mmbaga BBO (1994) Predicting surgical compliance in a cohort of women with trichiasis. International Ophthalmology 18, 105 109.
  • World Health Organization (1996) Future Approaches to Trachoma Control. Report of a global scientific meeting, Geneva. 17–20 June 1996. WHO, Geneva.
  • World Health Organization (1997) Global Initiative for the Elimination of Avoidable Blindness. WHO/PBL/97.61.WHO, Geneva.
  • World Health Organization (1998) Global elimination of blinding trachoma.WHO, Geneva. Resolution WHA 51.11 adopted by the World Health Assembly 16 May 1998. Text at http://www.who.int/pbd/trachomawha51-e.htm.