METHODS

Localisation and study design

The city of Colima, capital of the State of Colima, Mexico, is located in the Mexican Pacific coast at 19° 15` N and 103° 43` W. It has a population of 120 780 habitants, a semi-humid and warm climate, an annual average temperature of 25.2°C, and annual average pluvial precipitation of 1010 ml. Its altitude is 490 m above sea level.¹⁶ The study was done in the west end of the city where a project named “A New Initiative” (Programa UNI-Colima) has been developing for more than eight years and is sponsored by the University of Colima and the WK Kellogg Foundation. This project consists of a series of activities directed to encourage the active participation of the community in the resolution of their health problems. These activities, in which the natural leaders of the community get involved, as well as the university, some municipal and state health authorities, are in accordance with the recommended outlines by the Local Health Systems.^17,18 The UNI programme covers an approximate surface of 20 km², with an estimated population of 30 000 persons.

This was a randomised community trial, in which the observational units were 187 houses, randomly selected by means of a draw from between approximately 7000 houses in the west end of Colima. This area was chosen because of the homogeneous socioeconomic conditions of the habitants and the access to public services, in such a way, they are non-stratified variables. The 187 houses were grouped in four blocks, in accordance to type of received intervention: the first block (A), consisted of 46 houses that received only malathion ULV spraying, applied by the State Secretary of Health. Block (B) was structured by 45 houses that did not receive any type of treatment (control group). Block (C) included 49 houses in which ULV were applied combined with an educational campaign and Block (D) with 47 houses where only an educational campaign was used.

Sampling methods

The sampling methods consisted of an entomological survey for surveillance of the Aedes aegypti, as recommended by the Pan American Health Organisation⁴; in which an operator spent about 15 minutes for each house. Inspection was done in rooms, bathrooms and patios, looking for any type of receptacle with water, with the help of a hand lantern. Only wet containers were registered and the presence of at least one larva, in any instance, was considered positive in a nominal form. Of the positive containers, a sample of larvae was taken to determine the specie and to confirm the presence of Aedes aegypti. Simultaneously to the entomological sampling, a survey was done regarding knowledge, attitude and practices relating to dengue and the vector mosquito, in accordance with the outlines recommended by Rosembaun.¹³ From this survey, the indicator of knowledge, attitude and practices was obtained (KAP), which was expressed in the continuous interval scale from 1 to 12. The environmental temperature was measured in centigrade degrees (°C) each hour, by means of an environmental thermometer at the university's observatory, which is located close to 2 km from the studied area. Based on this measurement, the average weekly temperature was determined during the studied period; which started in the warm rainy season (September to November 1998) and concluded in the cold dry season (February to April 1999).

Treatments

The educational campaign (EC) consisted of a series of visits house by house, done by college students of the University of Colima. The habitants, principally housewives, were offered a talk about the importance of dengue, ways to prevent it and biological characteristics of the Aedes aegypti. Such talks were re-enforced with group meetings in which a dengue video was shown and a sociodrama was presented in the kindergarten of the area. In addition, in each visit the message was supported with a small gift, consisting of sweets, stickers and calendars related to dengue and the Aedes mosquito. The community leaders facilitated the entrance to the houses, and that way an average of three visits per house was achieved.

The insecticide application was done by personnel of the State Health Secretary of Colima by means of spatial spraying of malathion to 95% at ULV (drop size 3 μ). A “London Fog” equipment was mounted on a pick up type truck, moving at 16 km/h, allowing an approximate dose of 120 mg/ha of insecticide. The ULV were applied following the streets of block A and C in cycles of seven consecutive days during October and November of 1998, and repeated in January and February 1999. At the same time, a group of technicians of the same health department, were in charge of application of granulated temefos (Abate) in each container with water that had been found in the mentioned blocks, at a doses of 1 ppm. The B and D blocks, even though they did not receive any ULV nor temefos in a systematic way, were occasionally subjected to those treatments when persons reported an increased nuisance by mosquito bites. In C and D blocks, where EC was applied, the population was encouraged to also use some forms of chemical or biological control, such as intradomestic sprinklers of piretroids, use of temefos, larvivorous fishes or Bacillus thuringensis; but always under their own initiative. By the conclusion of the study, confirmation had been done that the houses that should have received adequate intervention had received such. Houses were eliminated from the study if the habitants had not received adequate intervention or the proper management had not been carried out.

Statistical analysis

As well as the traditional larval indices, House, Container or Breteau index; the number of positive containers with larva by house (C+/H) were used as the dependent variable in an interval continual scale, with its mean and standard deviation, in such a way that this variable could be analysed with the statistical parametric method, such as Student's t test and the one way, or two way analysis of variance. With the purpose of exploring the correlation of C+/H with the temperature and the KAP index, an uni-variant unconditional logistic regression was used, where C+/H is expressed as a dichotomous variable, the previous weekly temperature at sampling, in a continuous scale of proportion, meanwhile KAP was transformed in a categorical multiple variable. The changes in the KAP index before and after the intervention and between the different blocks, expressed as a continual variable interval, were also analysed with the Student's t test and analysis of variance. For the statistical analysis, the program Statistica for Windows 1997 was used.

RESULTS

Two hundred and fourteen houses were originally visited, but 27 were discarded as mentioned previously, so finally completed data were obtained for for 187 houses. Table 1 shows the observed changes in the number of positive containers by house (C+/H) in each block and by global level, before and after the intervention. The basal House Index was 54% and after treatment 41.17%; Basal Container Index was 35% and after treatment 31.2%; while the basal Breteau Index was 97.3 and after treatment 70.

The most frequently infested containers with larvae in both seasons were badly covered subterranean water cisterns, followed by drums and courtyards built in washbasins (“piletas”); while in rainy season, larvae were also were found in tyres, ceramic and miscellaneous pots with water. Other types of containers were rarely infested (flower vases, drainage, roof tanks, etc). Ninety five larvae were examined, of which 82 corresponded to Aedes aegypti, (Linneo); seven to Culex pipiens q. (Say), three to Aedes epactius (Dyar and Knab) and three to Culex sp.

In table 2 changes obtained in C+/H by the end of each intervention are shown with each type of the treatments, as well as the analysis of such changes by means of a two way analysis of variance. The average temperature during the first part of the study (rainy warm temperature) was 24.61°C (95% CI 24.4 to 24.7), while the second part, in the dry season, was of 23.92°C (95% CI 23.1 to 24.3). The logistic regression between C+/ H and temperature did not show any significant association (χ=2.1, p>0.5, OR=0.3; 95% CI 0.13 to 1.67). The pluvial precipitation during the period September to December 1998 was 434 ml, while for January to April 1999 period was only 4 ml.

The KAP index did not show any significant changes after the intervention of the total of houses (basal= 5.69; 95% CI 5.07 to 6.41 v 5.89; 95% CI 5.45 to 6.23 after intervention). The one way analysis of variance for the difference of basal KAP and after treatment in the four blocks also did not show any significant differences (F=1.14, p>0.1), while the correlation between KAP y C+/ H showed χ=1.73, p>0.5; OR=0.87, 95% CI 0.23 to 2.34.

DISCUSSION

The great quantity of positive containers for larvae by house and infested houses found in this study differs notably from that informed by the Health Authorities of Colima, who rarely report values over 0.3 by house. This discrepancy is not new. Natham¹⁹ has noted that usually there exists great differences in the larval indices from when data are obtained in a simultaneous manner from an operative personnel and compared with ones recorded in works of investigation. Furthermore, the use of flashlight, originally proposed in the entomological survey and later abandoned by the official programmes, could have increased the sampling sensibility in this study, as the detection of small larvae in certain places, such as underground tanks and other very shadowed places, frequently positive to larvae, could easily go unnoticed to the eye of the most experienced technician without the help of a flashlight.

The increased frequency of C+/H permitted a more tough statistical parametric analysis, because of the probability of obtaining positive containers, with the number of examined houses, were sufficient to normalise a distribution curve with no need to transform the data. Despite how reasonable this assumption could be, almost all the studies relating to the analysis of larval indices are based on non-parametric statistics, assuming that the distribution of larva containers is non-Gaussian.¹² In this manner, assuming that there is a normalised distribution of the C+/H, analysis of variance by one and two ways is used, to compare the effects of the different type of treatments and for measurements before and after application. The total values of C+/H were sensitively reduced in the second phase of the study, which could be attributed at first glance to the climatic changes. Yet analysis of variance clearly shows that the EC has a much more accentuated effect than the use of chemicals and the control houses over the number of C+/H and because of this, they are the principal factors that reduced the larval index. In table 1 the basal values of C+/H were shown to be significantly higher in the D block (which received only EC intervention), which could lead to assuming that in that block the values of Δ C+/H after the treatment could be less, even though, the covariance analysis proved that the reduction of this parameter is even more significant after the EC (as can be appreciated in table 2).

Also in table 2 it can be appreciated that two way analysis of variance showed that the effect of the EC alone had a more significant reduction of Δ C+/H in relation to the use of ULV and to the control group, while the same analysis showed that the interaction between the EC and the use of malathion ULV reduces sensitively the effect of the first, which can be translated to a negative interaction, suggesting that the use of ULV can reduce the beneficial effect of EC. This phenomenon could be explained with the feeling of false security that is generated by the fumigating trucks passing, which makes people reduce the activities to eliminate breeding places. The presented data back the findings by Castle,²⁰ who found that the use of aerial malathion ULV do not produce significant reductions in the populations of Aedes aegypti, and are also congruent with the predictions done with mathematical simulated models that show a very brief and discrete effect of the ULV over the adult populations of Aedes aegypti.²¹ In this manner, the study confirms, in a quantitative manner, that an intensive EC, personalised and based on the active participation of the community leaders, is much more effective than the traditional applications of chemicals in a vertical form for the control of Aedes aegypti and the dengue transmission.

The reductions in larval indices obtained here are much more noticeable that the ones found by other similar studies. For example, Lloyd¹¹ in Yucatan, Mexico did not find any reduction in such indices, but only a lesser increase in the same after an EC, in comparison with a control group of houses. Similarly, Rosenbaum,¹³ in a study in Trinidad and Tobago, did not find any significant differences in the larvae indices after applying an intense campaign of community participation. It is possible that in the present work, the previous visit, done by the UNI programme to the community, has enhanced the EC impact by facilitating the admission to the homes and the acceptance of the participating technicians in the study. Also, it is very probable that the use of permanent printed material, such as stickers and calendars, may have caused some kind of subliminal effect that might also have been favourable to the actions of the particpants to eliminate the breeding places of the mosquito. It is suggested that before programmes based in the community for the control of Aedes aegypti are implemented, they should be preceded by community actions in which different sectors are involved such as universities, local authorities, health institutions and civil organisations as has been emphasised by some authors.^17,22

In this study, the environmental temperature variations and the pluvial precipitation did not show any correlation with the changes of C+/H, which, apparently is in disagreement with that expected, as the temperature clearly influences the reproduction rate and the population density of mosquitoes.^6,23 One probable explanation for this discordance could be that in places like Colima, where temperature fluctuations are so discrete, the impact of this climatological variable may have not the magnitude of those of areas more temperate, where habitually the field studies have been done. Therefore, the populations of Aedes aegypti could show intrinsic changes independent of the climatic variables, as has recently been proposed.²⁴

The results of the KAP indicator did not show any significant changes or correlation with the number of C+/H at any time during the study, nor between the groups. These findings are similar to the ones reported by Rosenbaum¹³ and by Dias.⁹ It would seem that in the case of dengue, this sociological parameter is incapable of detecting attitude changes at a subconscious level that induces the activities of health self maintenance, even when there is not enough knowledge and that these surveys usually do not consider the cultural context of the community, which could play a very important part in the underlying changes in the mentality towards health problems in specific communities.²⁵ Because of this, it has recently been suggested that these kind of surveys could have a poor association with the EC outcomes¹⁴ so, perhaps the KAP surveys should be substituted by another type of tool that permits a better evaluation of the programmes against dengue and other transmissible illnesses, or at least limit their use for qualitative appraisal of the programmes.

In summary, this study shows that the use of EC house by house, preceded by an extensive work of inter-sectorial integration with the community has a clear effect on the reduction of the breeding places for Aedes aegypti and with that, the risks for transmission for dengue. The use of chemicals in the traditional manner (ULV), can delay the effect of such an EC, and for that reason, should reserved only for intense epidemical outbreaks.