Impact of annual dosing with ivermectin on progression of onchocercal visual field loss.

Impact of annual dosing with ivermectin on progression of onchocercal visual field loss* S.N. Cousens,1 A. Cassels-Brown,2 1. Murdoch,2 O.E. Babalola,3 D. Jatau,4 N.D.E. Alexander,2 J.E. Evans,2 P. Danboyi,4 A. Abiose,4 & B.R. Jones2 Reported are the results of a randomized, double-masked, placebo-controlled trial of annual ivermectin dosing in 34 rural communities, Kaduna State, northern Nigeria, where guinea savanna onchocerciasis is mesoendemic. A total of 939 individuals underwent Friedmann field analysis at the first examination and saw at least 19 spots in at least one eye. Of these, 636 (68%) completed a subsequent Friedmann field analysis 2-3 years later. The adjusted incidence rate ratio for the ivermectin group versus the placebo group was 0.64 (95% confidence interval (Cl): 0.42-0.98). There was some evidence that the impact of ivermectin was greatest among those who had received one dose of ivermectin. The majority of the deteriorations occurred in eyes that gave evidence of optic atrophy at the first examination. An analysis restricted to individuals with optic atrophy at baseline indicated a reduction of 45% in the incidence of visual field deterioration in the ivermectin group (95% Cl: 8-67%). Previous findings have shown that ivermectin has an impact on the incidence of optic atrophy. Our results indicate, for the first time, that ivermectin has a substantial impact on the progression of visual field loss among those with pre-existing optic atrophy. Introduction Onchocerciasis is one of the major causes of blind- ness. Two commonly reported ocular sequelae of onchocercal infection are optic nerve disease and chorioretinitis (1), both of which can cause dramatic reductions in the peripheral visual field, leaving only a small island of central vision (2). In such cases visual acuity may be largely unaffected until a late stage of the disease process. Ivermectin is now the drug of choice for the treatment of onchocerciasis, and annual dosing with it substantially reduces the incidence of optic nerve disease among infected individuals (3) and produces beneficial effects on other onchocercal eye lesions, including punctate keratitis (4) and iridocyclitis (5). Less is known, however, about the impact of ivermectin on visual function loss and, in particular, visual field loss. In this article we report the effect of * From: Kaduna-London Collaboration for Research in Onchocerciasis. I Department of Epidemiology and Population Sciences, London School of Hygiene and Tropical Medicine, Keppel St, London WC1 E 7HT, England. Requests for reprints should be sent to Mr Cousens at this address. 2 Ahmadu Bello University Teaching Hospital, Kaduna, Nigeria; and International Centre for Eye Health, Institute of Ophthalmol- ogy, London, England. 3 Ahmadu Bello University Teaching Hospital, Kaduna, Nigeria. 4National Eye Centre, Kaduna, Nigeria. Reprint No 5774 annual dosing with the drug on the progressive de- terioration of visual field loss in northern Nigeria. Methods The population studied and methods used in the trial have been described previously (3, 6), Briefly, a randomized, double-masked, placebo-controlled trial of ivermectin was conducted in 34 rural com- munities that were mesoendemic for guinea savanna onchocerciasis, in Kaduna State, northern Nigeria. A census was carried out on the communities and demographic data and pertinent information, such as previous use of diethylcarbamazine (DEC), was collected. Skin-snips were taken from villagers aged -5 years, who were then registered for inclusion in the trial and randomly assigned at the individual level using a blocked design to receive ivermectin or pla- cebo. Annual dosing with ivermectin/placebo was performed over a 3-year period. At the fourth an- nual round, all villagers received ivermectin. Stand- ard exclusion criteria were applied at each dosing. The trial participants underwent ophthalmic ex- amination prior to the first, third, and fourth annual dosings. At each examination all villages were vis- ited, and registered individuals were asked to report to a central location in their village. Each patient was screened for signs of optic nerve disease by a team of ophthalmic nurses who had received intensive training from two ophthalmologists (LM., O.E.B.) Bulletin of the World Health Organization, 1997, 75 (3): 229-236 © World Health Organization 1997 229 S.N. Cousens et al. before the start of the fieldwork. The screening ex- amination consisted of a series of visual function tests (including visual acuity, peripheral visual field assessment by counting fingers in all four quadrants, paracentral visual field assessment using a variety of tests, colour vision, and a test for red desaturation), applanation tonometry, examination of pupillary light response, and undilated fundoscopy to assess optic disc colour and cup: disc ratio. Individuals who failed the screening examination were referred to an ophthalmologist; the referral criteria included reduced visual acuity that did not improve to 6/9 with pinhole, failure of any of the other visual function tests, and intraocular pressure ¢'2.80 kPa (:21 mmHg), an abnormal pupillary light response, suspicion of disc pallor or a cup: disc ratio >0.5. In addition, individuals belonging to a computer- generated random sample of the trial population were also referred to an ophthalmologist, regardless of the outcome of the screening examination. The ophthalmologist examined the referred pa- tients in a mobile clinic equipped with a slit-lamp biomicroscope, applanation tonometer, and retinal camera. After the patient's pupillary light response had been checked, the individual concerned re- turned to the ophthalmic nurse for pupillary dilation, dark adaptation, and Friedmann mark I visual field analysis. Topical mydriatics (cyclopentolate hydro- chloride (1% (w/v)) and phenylephrine hydrochlo- ride (10% (w/v)) were instilled until the pupils were maximally dilated and nonreactive to light. The Friedmann mark I Analyser is a form of static perimetry in which the central 250 of visual field is measured at a distance of 33cm. A total of 46 illuminated targets (generated by electronic flash and modified by specified neutral density filters) are presented to the patient in patterns of two, three, or four at a time. The patient indicates how many points or "spots" were seen and localizes their position; the points are then recorded on a standard form with the corresponding filter value at which they were seen. This form of field analysis is less sensitive to some early visual field defects than some other methods, such as computer-assisted or Goldmann (7). The Friedmann Analyser performs relatively well at detecting visual fields with a central island or central defects (8). Since onchocercal defects essen- tially constrict the visual field with preservation of a central island (2), the Friedmann Analyser is suit- able for use in onchocercal populations. The mark II Friedmann Analyser tests 98 rather than 46 points, but the small advantage that this would provide (9) was not considered sufficient to warrant the addi- tional test time required, nor the increased potential for machine failure. Field loss, as measured by the Friedmann Analyser, may manifest itself as either a decrease in sensitivity or absolute loss of stimulus perception. In this article, we consider only absolute loss of stimulus perception. The Friedmann field analyses were performed in rooms darkened with blackout material. The pa- tient sat at the appropriate height and used the palm of one hand to occlude one eye. The nurse perform- ing the analysis explained the test in Hausa, the most widely spoken language in the area, and the macular threshold was determined. Individuals were encour- aged to maintain central fixation and were observed while the test was run at 0.4dB below the macular threshold, at a further 0.4dB below this value, and finally at maximum brightness, when necessary. The results of the test were noted on the standard Friedmann analysis form. At the first examination, the Friedmann test for each patient was performed by the ophthalmic nurse who had performed the screening examination; at the subsequent examina- tions, all Friedmann field tests were performed by one ophthalmic nurse. Following Friedmann field analysis, the patient returned to the mobile clinic. At the second and third examinations the ophthalmologist compared the Friedmann field result with any previous results using a quadrant-by-quadrant summary. If there were major incongruities, the Friedmann analysis was repeated. Patients then underwent posterior segment examination with slit-lamp biomicroscopy, direct and indirect ophthalmoscopy, and fundus fluoroscein angiography. Upon completion of the examination, patients were dosed with ivermectin/ placebo by a general physician or rural health officer. In order to identify individuals with visual field deterioration during follow-up, an initial analysis of Friedmann field changes was performed. The number of individuals showing an improvement was compared with the number showing a deterioration for each designated cut-off point (e.g. ¢12 spots lost or gained). We assumed that most apparent improvements were probably due to measurement variations rather than true improvements, and in or- der to maintain as high a specificity as possible, we chose as cut-off point the minimum number of spots lost/gained that produced a ratio of deteriorations: improvements :10: 1. The cut-off point thus identi- fied was a change of >19 spots seen/not seen in one eye (92 individuals with deteriorations versus 8 with improvements, 4 of whom received ivermectin). Be- low, "visual field deterioration" refers to the abso- lute loss of :19 of the 46 Friedmann spots in one eye during the period of follow-up. Only changes be- tween the first and last examination are considered here. Data were double-entered on microcomputers and analysed using SAS/PC and EGRET software WHO Bulletin OMS. Vol 75 1997230 Impact of Ivermectin on onchocercal visual field loss packages. Poisson regression was used to estimate the incidence ratios using person-time of follow-up as the denominator. The homogeneity of the rate ratio (ivermectin versus placebo) across different values of other variables was tested by fitting an interaction term and examining the likelihood ratio statistic (10). Results Follow-up At the first examination and dosing, 939 individuals registered in the trial underwent Friedmann field analysis and saw >19 spots in at least one eye (and thus were at risk of deterioration according to the definition chosen). Of these individuals, 636 (68%) completed a Friedmann field analysis at one or more of the follow-up examinations. The mean period of follow-up was 28.3 months (range, 17.0-38.0 months). Follow-up was similar in the two treatment groups (ivermectin = 314 individuals (66%), mean = 28.3 months; placebo = 322 individuals (69%), mean = 28.2 months). The rate of follow-up was not associated with age (P = 0.19), sex (P = 0.18), or microfilarial load at baseline (P = 0.18). Rates of follow-up were higher among individuals in the ran- dom sample (75% versus 63%, P < 0.001) and those who had reported having taken DEC (77% versus 60%, P < 0.001). Follow-up also varied according to the individual's Friedmann field status at baseline, being highest among those who had missed some, but not all spots in their worse eye (75%), compared with 64% among those who had seen all 46 spots and 68% among those who had missed them all (P = 0.04). Baseline characteristics of the treatment groups Table 1 shows the distribution of individuals and follow-up time by treatment group according to age at census, sex, microfilarial load at baseline, reported history of DEC consumption, membership of the random sample, and the results of the Friedmann field analysis at baseline. There were no major differ- ences between the two treatment groups with regard to any of these variables. All age groups were repre- sented, though the number of individuals aged over 64 years was comparatively small. There were more males (55%) than females. The majority of individu- als (71%) were skin-snip positive at baseline, 31% having loads >10 microfilarae (mf) per mg; very few, however, had loads >SOmf per mg (6%). About half (52%) reported having used DEC. A total of 270 individuals (42%) were from the random sample; 348 individuals (55%) saw all Friedmann spots in both eyes at baseline, while 67 individuals (11%) missed all 46 spots in one eye. Treatment with ivermectin/placebo Of the 636 individuals who completed follow-up, 255 (40%) underwent Friedmann field analysis only twice (at the start and end of follow-up), 286 (45%) three times, and 95 (15%) on four occasions. The mean number of analyses performed was similar in the two treatment groups (ivermectin = 2.73; pla- cebo = 2.76). Over the period of follow-up, 273 indi- viduals (43%) received three doses of ivermectin/ placebo between their first and final examinations, 287 (45%) two doses, 60 (9%) one dose, and 16 (3%) were excluded at each dosing round and thus were never dosed during the period of follow-up. The dis- tribution of the number of doses received by indi- viduals with follow-up in the two treatment groups was similar (Table 1). Visual field deteriorations A total of 92 individuals (14%) had "lost" at least 19 spots in one eye between the baseline and follow-up examinations. Of these individuals 19 had suffered bilateral deteriorations, while the other 73 had deteriorated unilaterally. Of these 92 individuals, 34 (37%) were in the ivermectin group and 58 (63%) in the placebo group (Table 1), giving an incidence ratio of 0.60 (95% confidence interval (CI) = 0.39- 0.92; P = 0.02) in the ivermectin group relative to the placebo group. The incidence of deteriorations tended to in- crease with age (P < 0.001), although there were relatively few instances among those aged 55-64 years, and was higher among males than females (P = 0.02) in both treatment groups (Table 1). Deteriorations appeared to be commoner among the skin-snip positive, particularly in the placebo group (P = 0.03). There was no evidence that reported DEC consumption prior to the initial examination was associated with the incidence of visual field deteriorations during the period of follow-up (P = 0.15). Deteriorations among members of the random sample (who had not necessarily failed the screening examination) were much less likely than among those who were not members of this sample (and who had failed the screening examination; P < 0.001). Visual field status at baseline was associated with rate of deterioration (P < 0.001), deteriorations tending to be more frequent among individuals who had missed spots at baseline. Although the incidence of deteriorations appeared to increase with the num- WHO Bulletin OMS. Vol 75 1997 231 S.N. Cousens et al. Table 1: Baseline characteristics, number of Friedmann field analyses undergone, number of doses of ivermectin/ placebo received, and incidence of deteriorations in Friedmann visual field analysis, by treatment group, among the study subjectsa Ivermectin group Placebo group No. with Incidence (per 100 No. with Incidence (per 100 n deterioration person-years) n deterioration person-years) Age (years) 5-14 24 1 1.80 23 0 0 15-24 45 1 0.96 38 0 0 25-34 67 4 2.52 68 14 8.88 35-44 76 9 4.92 88 18 8.76 45-54 57 11 8.28 53 13 10.20 55-64 29 3 4.56 36 3 3.60 >65 16 5 12.60 16 10 25.20 Sex Male 172 24 5.88 176 38 9.12 Female 142 10 3.00 146 20 5.88 Microfilarial load Negative 94 7 3.24 89 8 3.72 0.1-10.0 119 16 5.64 131 31 9.96 10.1-50.0 73 8 4.68 82 17 9.00 50.1-100.0 20 3 6.12 12 2 6.72 >100 2 0 0 4 0 0 Unknown 6 0 0 4 0 0 DECb taken in past No 156 14 3.96 149 22 6.48 Yes 155 20 5.40 173 36 8.64 Unknown 3 0 0 0 0 Undefined Random sample No 174 31 7.56 192 48 10.56 Yes 140 3 0.96 130 10 3.36 No. of spots missed in better eye at baseline 0 229 17 3.12 245 37 6.48 1-10 43 12 11.64 36 14 16.56 11-20 24 4 7.32 28 6 9.12 >21 18 1 2.16 10 1 4.32 No. of spots missed in worse eye at baseline 0 169 6 1.56 179 17 4.08 1-10 38 10 10.92 51 16 13.44 11-20 28 5 7.44 21 4 7.68 21-45 43 6 5.76 40 9 9.60 All 36 7 7.92 31 12 16.20 No. of Friedmann analyses performed 2 133 10 3.64 122 16 6.55 3 132 15 4.48 154 31 7.91 4 49 9 6.89 46 11 9.00 No. of doses of ivermectin/placebo received 0 7 1 6.72 9 1 5.28 1 36 1 1.32 24 8 15.00 2 137 9 3.12 150 23 7.44 3 134 23 6.36 139 26 6.96 Total 314 34 4.56 322 58 7.68 a A subsample was alsb examined 7-14 days after first dosing, so that some individuals underwent four Friedmann field analyses. b DEC: diethylcarbamazine. WHO Bulletin OMS. Vol 75 1997232 Impact of ivermectin on onchocercal visual field loss ber of Friedmann analyses performed, this trend was not statistically significant (P = 0.12). None of these factors appeared to confound the association be- tween treatment group and incidence of visual field deterioration. When each variable was entered individually into a Poisson regression model containing a term for treatment group, the rate ratio for ivermectin versus placebo varied between 0.59 (number of spots missed by worse eye at baseline) and 0.63 (member- ship of random sample). The impact of ivermectin could vary with age (rate ratio = 0.47, <45 years; rate ratio = 0.93, 45-64 years; rate ratio = 0.50, >65 years), but this variation was not statistically signifi- cant (P = 0.35). Also, there was no evidence that the impact of ivermectin on the incidence of visual field deteriorations varied with any of the following: sex (P = 0.50); microfilarial load (P = 0.84); DEC (P = 0.92); random sample membership (P = 0.14); visual field status at baseline (P = 0.70); and number of Friedmann field analyses performed (P = 0.84). The incidence of visual field deterioration was not associated with the number of doses of ivermectin/placebo received (P > 0.50; Table 1). However, when individuals who had not received any doses of ivermectin/placebo were excluded there was some evidence that the impact of ivermectin varied according to the number of doses received, the impact appearing to be greater among those re- ceiving one or two doses of ivermectin/placebo than among those receiving three doses (rate ratios: 0.09, 0.42 and 0.92, resp.; P = 0.02). A Poisson regression model was fitted to esti- mate the impact of ivermectin on the incidence of visual field deteriorations, adjusting for age, sex, microfilarial load, reported use of DEC, membership of the random sample, visual field status at baseline (worse eye), number of Friedmann field analyses performed, and number of doses of ivermectin pla- cebo received. Treatment with ivermectin was asso- ciated with a 36% reduction in the incidence of visual field deterioration (incidence rate ratio = 0.64; 95% CI = 0.42-0.98; P = 0.04). To investigate the etiology of the observed deteriorations, we reviewed the records of all 92 individuals with visual field loss to determine the prevalence among them of media opacities, chorioretinal, and optic nerve pathologies at the end of follow-up. A total of 24 individuals had evidence of media opacities, mostly in the lens, that could have restricted visual function (13 ivermectin, 11 pla- cebo). Of those individuals, 21 also had evidence of optic nerve and/or chorioretinal pathology. The distribution of optic nerve and choriore- tinal disease among the 92 individuals with visual field loss is shown in Table 2. More than 90% had Table 2: Distribution of optic nerve and retinal patho- logies at the end of follow-up among 92 individuals with visual field deteriorations, by treatment group No. of individuals Ivermectin Placebo Pathology group group Total Optic nerve only 15 (44)a 24 (41) 39 (42) Retina only 1 (3) 0 (0) 1 (1) Optic nerve and retina 14 (41) 33 (57) 47 (51) Neither pathology 4 (11) 1 (2) 5 (5) Total 34 (100) 58 (100) 92 (100) a Figures in parentheses are percentages. clinical evidence of optic nerve pathology, while just over half had clinical evidence of chorioretinal pa- thology. More than 40% had optic nerve pathol- ogy in the absence of chorioretinal pathology, while only one individual had chorioretinal pathology in the absence of optic nerve pathology. Discussion Our results suggest that annual dosing with ivermectin reduces by 36% the occurrence of visual field deterioration, which is similar to our previous finding that such dosing reduces the incidence of optic nerve disease by about 35% (3). This similarity is, perhaps, not surprising since optic nerve disease typically results in visual field loss. Our definition of an incident case of optic nerve disease was an individual with one or both eyes having onset of optic disc pallor associated with a deterioration in visual function and/or pupillary light response (these cases therefore constitute cases of optic atrophy). One approach to identifying visual function deterioration is Friedmann field analysis. Thus, it is therefore plausible that both of these out- come measures reflect the same deteriorations, in the same eyes, in the same individuals. This was, however, not the case. Of the 92 individuals we iden- tified with visual field deterioration, only 14 (15%) were considered to have suffered onset of optic nerve disease (optic atrophy) during the follow-up period. Furthermore, among these 92 individuals, 111 eyes were identified as having suffered visual field deterioration, and of the latter eyes, only 7 (6%) had developed optic nerve disease as defined above. The apparent inconsistency between the data for individuals and those for eyes arises because some individuals experienced onset of optic atrophy in one eye and visual field deterioration in the other. WHO Bulletin OMS. Vol 75 1997 233 S.N. Cousens et al. The majority of the 111 eyes with visual field deterioration (75%) already had evidence of optic atrophy at the baseline examination. Of the 13 random sample individuals with deteriorations, 12 (92%) failed the nurse's screening examination at baseline. The one individual who was not identified at baseline suffered a traumatic cataract during follow-up. Thus, while our previous findings (3) revealed that ivermectin had an impact on the inci- dence of optic atrophy, our present results suggest that it has also an impact on the progression of pre- existing disease. An analysis restricted to individuals with optic atrophy at baseline confirms the above impression (Table 3). Among such individuals ivermectin was associated with a 52% reduction in the incidence of visual field deteriorations (P = 0.004). There was no strong evidence to suggest that the impact varied with the number of doses received (1 dose, rate ratio = 0.27; 2 doses, rate ratio = 0.38; 3 doses, rate ratio = 0.63; P = 0.24). After adjustment for age, sex, microfilarial load at baseline, and visual field in the worse eye, the estimated reduction in the inci- dence of visual field loss in those receiving ivermec- tin was 45% (95% CI = 8-67%). Our data indicate that optic nerve pathology, alone or in combination with chorioretinal pathol- ogy, is an important cause of visual field loss in the study population; chorioretinal pathology in the ab- sence of optic nerve pathology does not appear to be important in this respect. This finding should, however, be interpreted with caution. The original ophthalmic nurses' screening examination, which identified individuals (other than those in the ran- dom sample) for Friedmann field analysis, was de- signed to identify those with optic nerve disease rather than chorioretinal pathology. Individuals with the latter pathology, in the absence of visual function defects at first examination, will in general have been excluded from our sample; however, those with chorioretinal pathology and visual function defects were included. We have previously reported that the impact of ivermectin on the incidence of optic nerve disease Table 3: Incidence of visual field deterioration in indi- viduals with optic nerve disease at baseline Ivermectin Placebo group group No. of individuals 118 109 Person-months of follow-up 3421 3100 No. of deteriorations 24 45 Rate ratio for ivermectin versus placebo 0.48 (0.29, 0.79)a a Figures in parentheses are the 95% confidence interval. appears to be largely confined to individuals with microfilarial loads >lOmf per mg. The incidence of visual field deterioration by microfilarial load shown in Table 1 suggests that ivermectin has little or no impact on such deterioration among skin-snip- negative individuals (rate ratio = 0.85), but may do so among those with loads <lOmf per mg. In Sierra Leone this threshold density had previously been suggested as being necessary for the occurrence of acute skin changes (11). Among individuals with loads -10mf per mg or >lOmf per mg, ivermectin reduced by 43% the rate of visual field deterioration. Such an interpretation of the data should be treated very cautiously, however, since small changes in the values could radically alter the observed pat- tern; had we, for example, observed two fewer deteriorations in skin-snip-negative individuals re- ceiving ivermectin, the rate ratio among such individuals would have been 0.61. There was no sta- tistically significant interaction between the impact of ivermectin and microfilarial load. Our results contain two apparent anomalies which may be related. First, we detected only six individuals aged 55-64 years with visual field deteriorations; in view of the pattern of deterio- rations in the other age groups, three or four times as many might have been expected. We have no expla- nation for this shortfall. Second, the impact of a sin- gle dose of ivermectin appears to be greater than that of two or three doses. The number of doses of ivermectin/placebo received was associated with age and sex - older individuals tending to receive more doses (P < 0.001) and 51% of males receiving all three doses compared with only 32% of women (P < 0.001). The number of doses received did not appear to vary with initial microfilarial load - 89% of skin-snip- negative individuals receiving two or three doses of ivermectin/placebo compared with 88% of skin-snip positives. The more doses of ivermectin/placebo received, the more likely the individual reported having used DEC (P < 0.0001), the less likely the individual was included in the random sample (P = 0.001), and the less likely the individual was to have seen all spots at baseline (P = 0.005). We did not skin-snip all individuals annually, since this procedure appeared to reduce participa- tion in the trial, and we aimed to achieve the highest level of ophthalmic follow-up possible. It is highly probable, however, that the first dose of ivermectin produced the greatest reduction in microfilarial load, with subsequent doses producing smaller reductions, as observed in other trials (12-15). In addition, Alley et al. (15) have reported that microfilarial loads can remain at around half their pre-treatment level for up to 4 years after a single dose of ivermectin. If the WHO Bulletin OMS. Vol 75 1997234 Impact of ivermectin on onchocercal visual field loss rate of visual field deterioration is associated with microfilarial load, over a 3-year period a single dose might be almost as effective as annual doses. Such a mechanism does not, however, explain why a single dose should appear more effective than two or three doses. A possible explanation for the observed interac- tion is that number of doses acts as a proxy for an- other variable. If, for example, older individuals are more likely than younger individuals to suffer non- onchocercal visual field loss (e.g. because of glau- coma or dense cataract), ivermectin would be expected to have less overall impact among older age groups - the very groups that received more doses of ivermectin/placebo. The plausibility of such an explanation is weakened, however, by our failure to demonstrate any direct interaction between ivermectin and age, sex, or any of the other variables considered. A total of 21 of the individuals with visual field deteriorations had cataracts at the end of follow-up (12 in the ivermectin group and 9 in the placebo group). Excluding such individuals from the analysis gives an estimated rate ratio of 0.49 (95% CI = 0.29, 0.81) for ivermectin versus placebo. The interaction between ivermectin and number of doses of ivermectin/placebo received is then no longer statistically significant (P = 0.10). The estimated rate ratios for one, two, and three doses, respectively, are 0.11, 0.35, and 0.74. Similarly, if only individuals with optic atrophy at baseline are considered, there is no strong evidence that the impact of ivermectin varies with the number of doses (P = 0.23). In defining visual field deterioration, we chose a largely arbitrary cut-off point (:19 spots lost in one eye). The corresponds to a major change in visual function. A strict cut-off point was chosen to ensure a high specificity for our outcome measure. Indi- rectly, the specificity can be assessed by examin- ing the incidence of deteriorations according to the number of Friedmann field analyses undergone by individuals. If the outcome lacked specificity, the number of deteriorations might have been expected to decrease with the number of investigations as the patient's performance improved. There was no evi- dence that the incidence of visual field deteriorations was lower among individuals who underwent four analyses rather than only two (Table 1). An alterna- tive indirect means of assessing specificity is to exam- ine the effect of cut-off point, and for this purpose we examined the following: -11 spots lost: -16 spots lost; and ¢21 spots lost. At these cut-offs the ratio of deterioration in the ivermectin group to that in the placebo group was 0.87, 0.72, and 0.49, respectively (at ¢19 spots the ratio was 0.59). These ratios indi- cate that the specificity increased with the cut-off point. Crude analysis of these cut-off points indicate that the statistical significance of the impact of ivermectin increased as the cut-off criterion became stricter (P = 0.38, P = 0.07, P = 0.01, P = 0.002). In summary, our findings indicate that annual dosing with ivermectin substantially reduces the rate of visual field loss in guinea savanna communities that are mesoendemic for onchocerciasis. As far as we are aware, this is the first report that ivermectin has an impact on visual function, rather than eye lesions; we have previously found that it has an im- pact on the onset of onchocercal optic atrophy. In the current study, the majority of cases of visual field loss that were identified occurred in eyes with optic atrophy at baseline. Thus, ivermectin appears to pre- vent or delay the onset of optic atrophy and subse- quently slow the progression of the disease, as measured by visual field loss. Acknowledgements We thank the ophthalmic nurses, Mrs I. Bolarin, Mr S. Labwane, Mr B. Lawal, Mr I. Liman, and Ms J. Onyema; Dr H. Yahaya and Dr N. Ngozi; and Dr W.G. Gregory and Mr J. Adebote; the rural health officers Mr A. Agwu, Mr B. Asha, Mr G. Audu, Mr R. Dakare, and Mr M.S. Moham- med; S. ldakwo, I. Nuhu, M. Zakari, and other members of the research team; the Director and staff of the Nigerian Institute for Trypanosomiasis Research; and the Kaduna State Ministry of Health, the local government authorities, and the villagers in the trial communities. This study was supported by WHO/UNDP/World Bank Special Programme for Research and Training in Tropical Diseases (TDR) (Project No. 870456, 880040, and 910553); the Leverhulme Trust (IM., A.C.-B., N.D.E.A., JE. B.R.J.); and Sightsavers Royal Common- wealth Society for the Blind. A Director's Initiative Grant from TDR (No. 870122) and a grant from the British Coun- cil for the Prevention of Blindness funded Dr S. Hussein in the preliminary survey and project design study. Resume Impact de I'administration annuelle d'ivermectine sur l'evolution de la perte de champ visuel due a l'onchocercose Cet article decrit l'impact de l'ivermectine en traite- ment annuel sur la perte de champ visuel, deter- min6 lors d'un essai randomis6 en double aveugle contre placebo. L'essai a ete realis6 dans 34 communautes rurales de l'Etat de Kaduna, dans le nord du Nig6ria, ou l'onchocercose de savane guineenne est mesoend6mique. L'exploration du champ visuel paracentral a ete r6alis6e a l'aide d'un appareil de Friedmann mark I au d6but de l'essai puis au bout de 2 et 3 ans de suivi. Une perte WHO Bulletin OMS. Vol 75 1997 235 S.N. Cousens et al. absolue de .19 points sur un ceil pendant la periode de suivi a ete prise comme seuil de deterioration du champ visuel. Lors de l'examen initial, 939 personnes ont vu .19 points avec au moins un ceil et ont pu etre incluses dans l'etude; l'examen du champ visuel a 6te repete pendant le suivi chez 636 d'entre elles (68%). Les taux de suivi ont ete similaires dans le groupe traite par l'ivermectine et dans le groupe placebo (dur6e moyenne = 28,3 mois), ces groupes etant bien apparies en ce qui concerne l'age, le sexe, la charge microfilarienne et l'etat initial du champ visuel. Au total, 92 personnes (34 dans le groupe ivermectine et 58 dans le groupe placebo) ont pr6sente une deterioration du champ visuel. Le rapport d'incidence corrige du groupe placebo etait de 0,64 (intervalle de confiance a 95% = 0,42- 0,98). 11 semble que l'impact de l'ivermectine soit plus grand chez les personnes qui ont requ une seule dose que chez celles qui ont recu trois doses. Une pathologie du nerf optique, seule ou associee a une pathologie chorio-retinienne, a ete observ6e chez plus de 90% des sujets ayant une deteriora- tion du champ visuel. Alors que I'association de ces deux pathologies 6tait courante chez ces sujets (51%), une pathologie chorio-retinienne isolee n'a ete observee que chez un seul sujet. La plupart des d6t6riorations (75%) ont ete observees chez des sujets ayant d6ja une atrophie optique au premier examen. Une analyse limitee a ces cas a indiqu6 une diminution de 45% de l'incidence de la det6rioration du champ visuel apres administration d'ivermectine (intervalle de confiance a 95% = 8- 67%). Nous avions deja rapport6 un impact de l'ivermectine sur l'incidence de l'atrophie optique dans cette meme population. Les resultats de la presente etude montrent que ce compose a aussi un impact sensible sur la progression de la perte de champ visuel chez les sujets deja atteints d'une atrophie optique. 11 semble donc que l'ivermectine empeche ou retarde I'apparition d'une telle atrophie et ralentisse l'6volution de la maladie, mesurable par examen du champ visuel. A notre connais- sance, c'est la premiere fois qu'un impact de l'ivermectine sur la fonction visuelle et non sur les lesions oculaires est rapport6. References 1. Sandford-Smith J. Eye diseases in hot climates. Bristol, Wright, 1987. 2. Thylefors B, Tonjum AM. Visual field defects in onchocerciasis. British joumal of ophthalmology, 1978, 62: 462-467. 3. Abiose A et al. Reduction in the incidence of optic nerve disease with annual ivermectin to control onchocerciasis. Lancet, 1993, 341: 130-134. 4. Whitworth JAG et al. Effects of repeated doses of ivermectin on ocular onchocerciasis: community- based trial in Sierra Leone. Lancet, 1991, 338: 1100- 1103. 5. Dadzie KY et al. Changes in ocular onchocerciasis four and twelve months after community-based treat- ment with ivermectin in a holoendemic onchocerciasis focus. Transactions of the Royal Society of Tropical Medicine and Hygiene, 1990, 84:103-108. 6. Abiose A et al. The distribution and aetiology of blindness and visual impairment in mesoendemic onchocercal communities, Kaduna State, Nigeria. British journal of ophthalmology, 1994, 78: 8-13. 7. Wild JM et al. The qualitative comparative analysis of the visual field using computer assisted, semi- automated and manual instrumentation: II. Statistical analysis. Documenta ophthalmologica, 1984, 58: 325-340. 8. Flanagan JG et al. The qualitative comparative an- alysis of visual field using computer assisted, semi- automated and manual instrumentation: Ill. Clinical analysis. Documenta ophthalmologica, 1984, 58: 341-350. 9. DeBoer RW et al. The Friedmann visual field analy- ser mark 11 - technical evaluation and clinical re- sults. Documenta ophthalmologica, 1982, 53: 331- 342. 10. Clayton D, Hills M. Statistical methods in epidemiol- ogy. Oxford, Oxford University Press, 1993: 239- 244. 11. Luty AJF et al. Immunological studies on oncho- cerciasis in Sierra Leone. I. Pretreatment baseline data. Tropical medicine and parasitology, 1990, 41: 371-375. 12. Lariviere M et al. Double-blind study of ivermectin and diethylcarbamazine in African onchocerciasis pa- tients with ocular involvement. Lancet, 1985, 2: 174- 177. 13. Taylor HR, Greene BM. The status of ivermectin in the treatment of human onchocerciasis. American journal of tropical medicine and hygiene, 1989, 41: 460-466. 14. Greene BM et al. A comparison of 6-, 12-, and 24- monthly dosing with ivermectin for the treatment of onchocerciasis. Journal of infectious diseases, 1991, 163: 376-380. 15. Alley ES et al. The impact of five years of annual ivermectin treatment on skin microfilarial loads in the onchocerciasis focus of Asubende, Ghana. Transac- tions of the Royal Society of Tropical Medicine and Hygiene, 1994, 88: 581-584. 236 WHO Bulletin OMS. Vol 75 1997