The onchocerciasis control programme in Mexico began treatment with ivermectin in 1989, initially providing treatment only to patients positive at nodulectomy and/or presenting with Mazzotti reaction (to oral diethylcarbamazine) who were residents from hyperendemic villages (microfilarial prevalence ≥ 60%). From 1991 to 1994, bi-annual treatment with ivermectin was extended to all eligible residents of mesoendemic villages (microfilarial prevalence between 20% and 59%), and to 25% of those residing in hypoendemic villages (microfilarial prevalence < 20%). From 1995 to 1997, the coverage in hypoendemic villages increased to 40% of all eligible residents. Since 1997, the national strategy has been to provide mass biannual treatments to every eligible resident from all the at-risk villages (from hypo- to hyperendemic villages), and to shift the emphasis from control to elimination.

In 1996, and before the present study was conducted, it had been assumed that the impact on transmission would be more evident in villages with high coverage of and compliance to ivermectin and nodulectomy; therefore three villages with such characteristics were selected for the study described here, namely: Las Golondrinas (92°39'17"W, 15°26'06"N, 920 m above sea level (masl)), Rosario Zacatonal (92°37'47"W, 15°27'25"N, 791 masl), and Nueva América (92°26'38"W, 15°17'08"N, 880 masl), which, prior to the introduction of ivermectin, had microfilarial prevalences of 69%, 79%, and 46% respectively 22. Las Golondrinas and Rosario Zacatonal are 7.0 km apart from each other and a coffee finca (Palestina) is located between 2.5 and 3.0 km from each village respectively; they all constitute locality I. The village of Nueva América is surrounded by three coffee fincas, namely, La Victoria, La Fortuna, and Santa Fe, which are located at a distance between 2 and 5 km from the village; they all constitute locality II. The study area is depicted in Fig. 1. The closest simuliid breeding sites were located approximately at 6.0 km from Las Golondrinas, 1.0 km from Rosario Zacatonal, 2.5 km from Palestina, 2.0 km from Nueva América, 2.0 km from La Victoria, 1.0 km from La Fortuna, and 0.5 km from Santa Fe, i.e. within the flight range of S. ochraceum s.l. 2324. Treatment is typically administered to the eligible resident populations of each village in January and June of each year. In 1996, the average ivermectin treatment coverage among the total population had been of about 80%. However, for the purposes of this study, the residents were not treated as usual in January and June, but immediately after parasitological examination, which took place in May and November to coincide with the commencement of each of the annual coffee seasons (see below). A census of each community was conducted 1016 at the start of each survey in 1997–1998 and 1998–1999, and eligible residents were treated just after the parasitological examination was completed.

The villages receive a biannual influx of temporary labourers (see below) who normally would be eligible for treatment but who, at the time of the study, were not on the records of the Mexican onchocerciasis elimination programme. Although the possibility cannot be excluded that the migrants might have been treated with ivermectin while at home (mainly in Guatemala, see below) it is also possible, and perhaps more likely, that they would have missed treatments at home because of being away in Mexico when treatment rounds were distributed. The temporary coffee workers would have therefore, been falling through the net in both countries.

Every year the cultivation of coffee includes two seasons, planting-clearing and harvesting. Planting-clearing is carried out from May through July and harvesting from November through February. The latter coincides roughly with the season during which onchocerciasis transmission has been deemed to be highest in neighbouring Guatemala 25. Coffee fincas' owners or administrative personnel hire the coffee labourers. These field labourers arrive en masse (most of them from Guatemala) usually by trucks, buses, or on foot, before the start of each coffee season. After completing each task required during the corresponding season, the labourers leave the coffee fincas in the same manner. As the coffee labourers are usually not the same individuals from season to season, the name, place of residence abroad, and date of birth of each labourer was recorded on a registration form and confirmed by comparison with the payroll lists or their identification papers.

The coffee labourers were informed about the study and consent was requested and obtained before proceeding with skin snipping and parasitological examination. Most (> 80%) of the coffee labourers were adult males (≥ 18 years old), and more than 85% of the adult male population in each of the four coffee fincas agreed to participate in the study. The remainder (<20%) was made up of accompanying females and children but less than 10% of these participated in the study. Because migration patterns could change over time, both coffee cycles of 1997 to 1998 and of 1998 to 1999 were investigated in order to assess any variation in the migrant human and biting vector populations. A migrant was defined as a person who had worked in the coffee plantation for a whole season. In the following season, this person was most likely absent, but occasionally, he could participate in the study again if returning as a temporary worker to the area. Less than 10% of the migrants returned to the same fincas where they had worked before.

All examined migrant coffee labourers who were skin mf positive were offered ivermectin. Treatment was not given to those migrants with a severe medical condition, or those who refused to comply, and it was administered only after parasitological examination had been completed (see next section), or before returning home. In 1999, after the present study was completed and the migrant workers stopped receiving treatment, the Mexican onchocerciasis elimination programme was immediately notified. The programme continued treating biannually the residents in the three villages, but in 2003, it selected 37 out of 39 formerly hyperendemic villages, and 13 formerly mesoendemic villages in Southern Chiapas to receive three-monthly ivermectin treatment (4 rounds per year). Migrant labourers working on the coffee plantations are at present targeted for treatment.

Using a corneoscleral (Holth type) 2 mm punch, four skin biopsies (two from the supra-scapular and two on the supra-iliac body regions) were taken from each participating individual in each village (residents) and coffee finca (migrants). Skin biopsies were incubated overnight in buffered saline, emerging mf counted and the snips weighed to estimate the intensity of infection per mg of skin 18. The parasitological examinations of both residents and migrants were conducted before the migrants commenced their work, i.e. in May and November of 1997, and in May and November of 1998.

Entomological surveys to estimate onchocercal infection levels in host-seeking S. ochraceum s.l. were carried out in the three villages and the four coffee fincas. Flies were collected from May to August 1997, during the coffee planting-clearing season, and from November 1997 to February 1998, during the harvesting season. The entomological sampling was repeated in the coffee seasons of 1998–1999 following the same scheme.

S. ochraceum s.l. females were collected when landing on consenting volunteers from the villages, with each collection team composed of an attractant and a collector as previously described 26. Daily entomological sampling, conducted for 15 days in four sites at each locality started at 07:00 hours and ended at 17:20 hours. Collections consisted of 20 min sampling units followed by 40 min breaks, and were conducted simultaneously in four sites for each community and coffee finca. All S. ochraceum s.l. specimens were dissected immediately after collection and during the rest period to determine parity 27. The abdomen, thorax, and head of each parous female were examined for all stages of O. volvulus larvae, which were identified according to descriptive statistics 28.

The point (crude) prevalence of skin infection with mf and the (arithmetic) mean number of mf/mg of skin were calculated for each study season and all examined residents in villages and migrant labourers in coffee fincas of localities I and II. Exact 95% confidence intervals were estimated for prevalence values 29. In addition, the community microfilarial load (CMFL: the geometric mean number of mf per skin snip in those aged ≥ 20 years 30) was also assessed for residents and migrants as the latter comprised mainly adults.

In onchocerciasis, and assuming random biting and lack of concentration of mf by the vector's saliva, the proportion of individuals who provide an infected blood meal to simuliid vectors has been approximated by the prevalence of skin-biopsy positives in the human population 3132. Based upon the specific Onchocerca-Simulium combination prevailing in the study area, the prevalence of infection in the thoraces of the vectors (which contain normally non-infectious L1 and L2 stages) is higher than the prevalence of infection in the heads (which contain only the infectious third stage larvae). Therefore, the prevalence of infection in the bodies of flies (heads plus thoraces) is the most sensitive indicator of parasite-vector contact, and thus it can be used effectively to monitor the presence of mf in untreated coffee fincas and villages subject to control. It has been proposed that assessment of the impact of any intervention on transmission be based on the total number of O. volvulus larvae (all stages) in bodies, and on detection of L3 in heads of flies 33. Therefore, in the present study, the prevalence of infection was calculated as the number of flies positive for any O. volvulus larval stage divided by the total number of parous flies examined; the associated (exact) 95% confidence intervals (95% CI) were determined 29, and the results were expressed per 1,000 parous flies 34.

Differences between proportions of infected flies between coffee seasons for each site and for each year, as well as before arrival (April and October), during the stay (May through July, and November to February), and after departure (August and February) of migrants in localities I and II were tested by means of the (one-tailed) Fisher's exact modification of the 2 × 2 chi-squared test 35. A p-value < 0.05 indicated statistical significance.

As shown in Table 1, the total number of migrants in coffee fincas for each season in 1997–1998 and 1998–1999 varied widely (from 35 to 850 individuals). In contrast, the numbers of the resident population ranged from 350 to 437 individuals. Thus, the ratio of migrants to residents varied from 0.1:1 in locality I during the planting season of 1997–1998 and that of harvesting in 1998–1999, to 2:1 in locality II during the harvesting season of 1997–1998, and that of planting-clearing of 1998 (Table 1). Overall, during the whole study 57% of the migrants came from the Guatemalan Department of Huehuetenango (North-western hypoendemic focus), while the remaining 43% came from the non-endemic Guatemalan Departments of San Marcos, and Petén. Only 0.01% came from other endemic areas of Southern Mexico. The migrant population working in the fincas during each coffee season mainly comprised newly hired people. The proportion of residents that were skin snipped ranged from 51 to 65% for locality I, and from 17 to 63% for locality II, whilst the equivalent proportions for migrants varied between 37 and 67%, and between 11 and 88%. There were infected migrants in all coffee fincas, their microfilarial prevalence ranging from 3% (95% CI = 2–6%) in locality II at the start (November) of the harvesting season of 1997–1998 (when they represented 71% of the total population at the time), to 23% (10–42%) in locality I at the beginning (May) of the planting season of 1997 (when they represented 9% of the total population). CMFL values for the migrants ranged from 0.03 to 0.71 mf/snip. The prevalence in the resident population of the villages in locality I ranged from 13% (9–18%) in 1997 to 4% (95% CI = 2–8%) in 1998, but apart from these values, none of the other figures were significantly different. In Las Golondrinas, treatment coverage ranged between 80 and 88% of the population. In locality II microfilarial prevalence varied between 14% (10–20%) in 1997 and 7% (2–16%) in 1998. In general the prevalence of infection in the migrant population was higher in locality I than in locality II, but this difference was significant only for the planting season of 1997, and the ratio of migrants to residents was higher in locality II than in locality I.

A total of 28,999 and 25,823 S. ochraceum s.l. females were collected, respectively in villages and fincas during the study (Table 3). The entomological data are summarized by locality in Table 2. The percentage of parous flies was significantly higher in locality I, 73.8% (95% CI = 73.3–74.2%) than in locality II, 68.2% (67.5–68.8%). However, the number of flies harbouring O. volvulus larvae of any stage per 1,000 parous flies was significantly lower in locality I, 0.8 (0.5–1.2) than in locality II, 2.3 (1.5–3.2).

Table 3 disaggregates the entomological data by site and coffee season. For reasons of sample size the data for the three fincas of locality II (namely, Victoria, Santa Fé and Fortuna) are combined. During the years 1997–1998, the proportion of parous flies was significantly higher for all sites (all p-values << 0.01) in the planting-clearing season (May through July) than in the harvesting season (November through February). Accordingly, the number of infected S.ochraceum s.l. per 1,000 parous flies tended also to be higher during the former than during the latter coffee season, but this was statistically significant only in Finca Palestina (p = 0.012) of locality I and Fincas Victoria, Santa Fé, and Fortuna combined (p = 0.021) of locality II; the values in the villages of Las Golondrinas and Rosario Zacatonal are suggestive of the same trend but only of borderline significance (p = 0.05). The proportion of infected flies in the coffee fincas of locality II was significantly higher than that of the finca in locality I (p = 0.02) during May-July 1997, and also (p = 0.02) during November 1997–February 1998. Due to lack of entomological data for one of the fincas during the planting season of 1998 a similar comparison was not possible, and although suggestive of the same trend, the difference for the harvesting season of 1998–1999 was not significant

The proportion of infected parous S. ochraceum s.l. flies in population samples was also analyzed within coffee seasons, i.e., before the arrival of migrants (April and October), during the stay of migrants (planting-clearing in May through July, and harvesting in November through February), and after departure of migrants (August and February), and compared both between and within localities I and II. The data are summarized in Fig. 2. There was a significant difference (p = 0.0069) between proportions of positive flies between localities when all time periods were combined (Fig. 2D), with locality II exhibiting a higher infection rate (see also Table 2). This was attributable to the higher prevalence of flies with O. volvulus larvae found during the stay of the temporary workers (Fig. 2B). Overall there was a significant difference (p = 0.0126) between the during and after periods (Fig. 2G), with the proportion of flies infected during the stay of the temporary workers being higher than that after their departure. Although the proportion of positive flies during the stay period was higher than that in the before period (particularly for locality II), the difference was not significant (p = 0.45). The proportions of positive flies after departure of the workers were always lower than those during their stay (Figs. 2E for locality I and 2F for locality II), but this difference was statistically significant (p = 0.0126) only when all sites were combined (Fig. 2G).