Participation in three consecutive mass drug administrations in Leogane, Haiti
Participations lors de trois administrations massives de médicament à Leogane en Haïti
Participación en tres administraciones masivas de medicamentos consecutivas en Leogane, Haiti
Summary
enObjectives In the global effort to eliminate lymphatic filariasis, mass drug administrations (MDAs) are organised annually. The success of this strategy depends on achieving high levels of drug coverage, which reduce the number of persons with circulating microfilariae and consequently transmission. Persons who consistently fail to participate in MDAs represent a potential threat to the goal of filariasis elimination. We wanted to know the drug coverage, the proportion of persons who were systematically non-compliant and factors associated with this behaviour.
Methods We conducted three surveys following the third annual MDA of a filariasis elimination program in Leogane, Haiti: (1) a total population survey to determine coverage; (2) an adult survey to determine non-compliance and associated factors and (3) an urban survey to make a rural–urban comparison.
Results During the third MDA, the overall surveyed coverage was 78.5% [95% confidence interval (CI) 74.4–82.6] A survey among adult population showed coverage estimates for persons >14 years old of 59.4% (95% CI 52.0–66.7), 61.0% (95% CI 55.0–67.4) and 67.3% (95% CI 60.5–74.0), for the first, second and third MDA respectively. The coverage in rural areas (78.3%) was significantly higher than in urban areas (68.3%, P < 0.05). Of the population >14 years of age, 18% never took the drugs during any of three MDAs. These persons did not differ significantly from MDA participants by age, gender or other characteristics that we assessed.
Conclusion More research is needed to identify characteristics of systematically non-compliant persons in order to refine health education messages and improve distribution strategies to increase drug coverage.
Abstract
frObjectifs Dans l'effort global d’élimination de la filariose lymphatique, des administrations massives de médicament ont été organisées annuellement. Le succès de cette stratégie dépend de l'atteinte d'un haut niveau de l’étendue de couverture médicamenteuse, ce qui réduira le nombre de personnes porteuses de microfilaires circulant et par conséquent réduira la transmission de la maladie. Les personnes qui ne participant pas à ces administrations massives de médicaments représentent une menace potentielle pour le but de l’élimination de la filariose. Nous avons voulu connaître l’étendue de la couverture médicamenteuse, la proportion de personnes systématiquement non compliantes et les facteurs associés au comportement de non compliance.
Méthodes Nous avons mené trois enquêtes à la suite de la troisième administration de masse annuelle de médicament d'un programme d’élimination de la filariose à Leogane en Haïti: 1) enquête sur la population totale pour déterminer l’étendue de la couverture, 2) enquête chez les adultes pour déterminer la non compliance et les facteurs associés et 3) enquête urbaine pour une comparaison entre le milieu rurale et urbain.
Résultats Au cours de la 3ieme administration en masse de médicament, la couverture totale était de 78,5% (IC95%: 74,4–82,6). Une enquête dans lapopulation adulte a révèle une couverture chez les plus de 14 ans estimée respectivement à 59,4% (IC95%: 52,0–66,7), 61,0% (IC95%: 55,0–67,4) et 67,3% (IC95%: 60,5–74,0) pour la 1ere, 2ieme et 3ieme administration en masse de médicament. La couverture dans les zones rurales (78,3%) était significativement plus élevée que dans les zones urbaines (68,3%, p < 0,05). 18% des plus de 14 ans n'avaient jamais pris les médicaments quelle que soit l'année. Ces derniers n’étaient pas significativement différents des autres participants par l’âge, le sexe ou autres caracteristiques évaluées.
Conclusion Des études supplémentaires sont nécessaires pour identifier les caracteristiques des personnes systématiquement non compliantes afin d'affiner les messages d’éducation et d'améliorer les stratégies de distribution pour augmenter l’étendue de la couverture.
Abstract
esObjetivos Como parte del esfuerzo global para eliminar la filariasis linfática, anualmente se organizar administraciones masivas de medicamentos (AMM). El éxito de esta estrategia depende del poder alcanzar niveles altos de cobertura de los medicamentos, que reduzcan el número de personas que circulan con microfilarias y por ende, la transmisión. Las personas que consistentemente fallan en participar de las AMM representan una amenaza potencial al objetivo de eliminar la filariasis. Queríamos averiguar cual era la cobertura de medicamentos, la proporción de personas que sistemáticamente no cumplían y los factores asociados a este comportamiento.
Métodos Se realizaron 3 encuestas tras la tercera AMM anual del programa para la eliminación de la filariasis, en Leogane, Haiti: 1/encuesta a la población total con el fin de determinar la cobertura 2/encuesta entre los adultos para determinar falta de cumplimiento y factores asociados y 3/encuesta urbana para establecer una comparación rural-urbana.
Resultados Durante la tercera AMM, la cobertura total entre los encuestados fue del 78.5% (95% intervalo de confianza (IC) 74.4–82.6). Una encuesta entre la población adulta mostró estimativos de cobertura para mayores de 14 años del 59.4% (95% IC 52.0–66.7), 61.0% (95% CI 55.0–67.4) y 67.3% (95% CI 60.5–74.0), para el primer, segundo y tercer AMM respectivamente. La cobertura en áreas rurales (78.3%) fue significativamente mayor que en áreas urbanas (68.3%, p < 0.05). De la población < 14 años, el 18% no tomó medicamentos durante ninguna de las tres AMM. Estas personas no eran significativamente diferentes de quienes sí participaban de las AMM ni en edad, género o cualquier otra característica evaluada.
Conclusiones Se requieren más estudios para identificar las características de personas sistemáticamente no cumplidoras, con el fin de redefinir los mensajes de educación en salud y para mejorar las estrategias de distribución que aumenten la cobertura de medicamentos.
Introduction
Lymphatic filariasis (LF), caused by infection with the mosquito-transmitted parasites Wuchereria bancrofti or Brugia malayi, is ranked by the World Health Organisation (WHO) as one of the leading causes of permanent disability worldwide (World Health Report 1995). In more than 80 countries, 120 million people are infected, while almost one-fifth of the world's population is at risk (Weekly Epidemiological Record 2001). WHO recommends that a single dose of diethylcarbamazine (DEC) or ivermectin, combined with albendazole, be given yearly to eliminate LF (Ottesen et al. 1999). The concept underlying the elimination strategy is to decrease the pool of microfilaria in humans by treating a large enough proportion of the endemic population for a sufficiently long time to break the transmission cycle. Though it is assumed that a high coverage for 5–10 years will be adequate to achieve this goal, little is known about the risk represented by persons who consistently refuse to take the medication and who can act as a reservoir for microfilaria.
Since October 2000, Hôpital St Croix (Leogane, Haiti) has organised an annual mass antifilarial drug administration (MDA) with DEC and albendazole in the plains of Leogane. In the first 3 years of the program, 74 000, 55 000 and 94 00 persons were treated resulting in a reported coverage of 68.5%, 53.0% and 86.6%, based on the estimated total population of 108 000. A probability survey conducted after the first MDA found a surveyed coverage estimate of 71.3% [95% confidence interval (CI) 66.7–75.9; Mathieu et al. 2003]. We conducted the present survey to verify the reported drug coverage during the third MDA, and to determine the proportion of people who never took the drugs during the three distributions; i.e. those who were systematically non-compliant. To improve the health education and drug distribution strategies, we also assessed factors that might have influenced compliance such as health education, socio-economic status (SES), education level and possible differences between urban and rural populations.
Materials and methods
The survey was conducted in the plains of Leogane Commune (Sections Dessources, Petite Riviere and Grande Riviere), including the city of Leogane. Each year, DEC and albendazole were distributed for 4–5 days at 70–140 distribution points located in private houses, churches, and public buildings. Except for the first MDA, drugs were also distributed to children while they were attending primary and secondary schools. An intensive filariasis health education program preceded each of the three distributions in 2000–2002 (Mathieu et al. 2004). The methodology used in this survey was tested by Mathieu et al. (2003), in which three survey methods were compared in terms of accuracy and the resources needed to conduct the survey. The cluster survey design used here gave results that were comparable with those from a cluster survey in which probability was proportional to estimated size, but required fewer resources.
Total population survey (MDA 3)
Forty distribution posts were randomly selected among the 126 rural and 14 urban posts used for the third MDA. Ten households near each of the selected distribution points were randomly chosen for interviews following a method previously tested and described (Mathieu et al. 2003). Briefly, the first house was chosen approximately 100 m from the distribution point in a randomly determined direction by spinning a bottle. A ‘next-nearest-house’ path was used to select the next nine houses as described by WHO for the Expanded Program on Immunisation (EPI method) (WHO 1991). Four teams, each consisting of a program staff member and a Leogane resident, administered two standardised questionnaires.
A coverage questionnaire was used to collect data regarding whether each household resident took antifilarial drugs during the last MDA. All household members living in the selected houses at the time of the drug distribution were included in this coverage survey, regardless of their age (total population survey). For small children and persons not present at the time of the survey, the questions about MDA participation were asked of an adult household member. The survey coverage is defined as the total number of persons taking the medication divided by the total number of surveyed persons.
Adult survey (MDA 1–3)
In each selected household, one adult (>14 years old), randomly chosen among the persons present during the drug distribution, was asked to answer a more detailed questionnaire. This questionnaire included questions about LF knowledge, health education messages, participation in the three consecutive MDAs and side effects after each of the MDAs, level of education, and possession of bicycle, motorbike or car. The later was used to assess SES. The questionnaire was composed of open-ended questions, but to facilitate data entry, questions had most likely responses listed, and a space for the interviewer to record other answers. If the selected person was not present, the household was skipped and not replaced.
Rural–urban comparison
We hypothesised that rural and urban coverage was different. As we knew that only few of the 14 urban distribution sites would be selected for the total population and adult surveys, we decided to survey all the remaining urban posts to enable us to do a rural–urban comparison. The selection of the houses and the questionnaires used were the same as described above.
Data analyses
All the data were analysed with SAS version 8.1 (SAS Institute Inc., Cary, NC, USA) and with SUDAAN (Bieler 2001). The data for the random sample of the entire Leogane Plains were analysed separately from the data derived from over-sampling the urban area. Comparisons of categorical data were made using Mantel Haenszel Chi square test.
Results
Total population survey (MDA 3)
In the total population survey, we collected information from 1767 persons living near 40 distribution posts. The median family size was 5.6 persons (range: 1–20), the median age was 24.7 years (range: 0–93 years) and 54.4% of the respondents were female. The surveyed coverage estimate for the total population in the third MDA was 78.5% (95% CI 74.4–82.6). Among persons > 14 years of age, the coverage estimate was 77.2% (95% CI 73.1–81.4). Men were more likely to have taken the drugs than were women (OR = 1.4, 95% CI 1.1–1.8). Primary school aged children had the highest coverage (94.2%, 95% CI 90.3–98.1) (Figure 1).
Age related coverage estimates and 95% confidence intervals, total population survey, third MDA, Haiti, Leogane. Data represent between 25 and 437 persons per age group.
Adult survey (MDA 1–3)
For the adult survey, 392 persons (>14 years of age) were interviewed. The median age of the respondents was 30 years (range: 14–83 years) and 69.8% were women. Fifty-nine per cent of the persons interviewed had primary schooling or less. Sixty-three per cent of the households owned a bicycle, while 4% of the households owned a motorbike and 6% a car.
The drug coverage estimates derived from this questionnaire for the first, second and third MDA were 59.4% (95% CI 52.0–66.7), 61.0% (95% CI 55.0–67.4), and 67.3% (95% CI 60.5–74.0), respectively. Overall, 41.9% (95% CI 35.1–48.8) took the drugs all 3 years; 32.1% (95% CI 26.2–38.0), 7.4% (95% CI 4.6–10.3), and 18.6% (95% CI 12.7–24.4) took the drugs twice, once and never, respectively (Table 1). The number of times a person participated in the MDA was not significantly associated with gender, age or SES.
| Participated in MDA | Original survey % (95% CI) n = 392 |
|---|---|
| Only MDA 1 | 5.8 (3.2–8.5) |
| Only MDA 2 | 1.6 (0.4–2.8) |
| Only MDA 3 | 9.8 (6.5–13.1) |
| Only MDA 1 and 2 | 6.9 (4.6–9.2) |
| Only MDA 1 and 3 | 4.5 (2.7–6.3) |
| Only MDA 2 and 3 | 10.9 (7.2–14.5) |
| 1, 2 and 3 | 41.9 (35.1–48.8) |
| Never | 18.6 (12.7–24.4) |
When asked why they did not take the drugs, almost one third of the non-compliant adults answered that they believed that the drugs make people ill (Table 2). This proportion was consistent for each of the three MDAs. Other reasons commonly given for not having taken the drugs were having been being absent during the MDA (17.5–23.1%) and having been sick (11.2–19.2%). The percentage of non-compliant persons who were not aware that drugs should be taken decreased with each round of MDA.
| Reason for not taking drugs | MDA 1 (%) n = 154 | MDA 2 (%) n = 149 | MDA 3 (%) n = 127 |
|---|---|---|---|
| ‘Makes me ill’ | 31.5 | 33.8 | 35.0 |
| Not present during MDA | 23.1 | 19.4 | 17.5 |
| Sick | 11.2 | 16.5 | 19.2 |
| Too busy | 8.4 | 10.1 | 5.8 |
| Pregnant | 7.7 | 6.5 | 10.8 |
| Didn't know I had to take the drugs | 7.7 | 2.2 | 0.8 |
| Other | 11.3 | 12.2 | 10.0 |
Forty-seven per cent (95% CI 36.1–57.0) of the respondents who took the drugs reported to have never suffered from adverse reactions. There was a decrease in adverse reactions between the first (50.2%) and second MDA (34.1 %, P < 0.05), but not between the second and the third MDA (28.9%, P > 0.05). Reporting adverse reactions was not associated with gender or age of the respondent (data not shown). People with higher SES, defined as possessing a motorbike or car, were more likely to have reported adverse reactions after the first MDA (OR 9.7, 95% CI 1.1–88.6) and the second MDA (OR 4.8, CI 1.0–23.2) than persons without these modes of transport. Persons who reported adverse reactions after the first MDA appeared less likely to have taken the drugs during the second MDA than persons who did not experience adverse reactions, but this difference was not statistically significant (OR 0.6, 95% CI 0.3–1.4). During the third MDA, the same trend was observed for persons reporting adverse reactions after the second MDA (OR 0.8, 95% CI 0.4–1.5).
Knowledge about LF was related to participation in MDA. Specifically, persons who knew that LF was transmitted by mosquitoes, believed that the tablets prevent disease or knew that once a year treatment was necessary were significantly more likely to have participated in each of the three MDAs than persons who did not remember these three health education messages (Table 3).
| Awareness of the following health education message | Never participated in MDA (%) n = 70 | Participated in one MDA (%) n = 65 | Participated in two MDAs (%) n = 84 | Participated in three MDAs (%) n = 158 |
|---|---|---|---|---|
| LF is caused by a mosquito | 45.7* | 56.9 | 61.9 | 70.2 |
| Tablets prevent LF | 47.1* | 47.7* | 63.1 | 79.1 |
| Tablet has to be taken once a year | 21.4* | 26.2* | 42.9 | 54.4 |
| Everyone has to take the medication | 66.7 | 63.1 | 75.6 | 73.6 |
- * P < 0.05 compared with people who participated in three MDAs.
Respondents were asked whether they had learned about LF through various channels of health education. Person-to-person communication and radio were cited most frequently as the means through which respondents learned about LF, 49.5% and 47.2% respectively (Table 4). Almost one quarter of the adult population mentioned hospital and LF program staff as a source of information. Video projections (13.3%), churches (9.8%) and schools (6.9%) were also identified as having provided information on LF, as well as family members (6.9%) and health workers (6.1%). Among printed materials, posters were the frequently reported source of information, mentioned by 11% of the respondents; banners were mentioned by 7.2% of the respondents. In general, systematic non-compliers were less exposed to health education in general than other persons (OR 0.32, 95% CI 0.15–0.70). However, when looking at each of the health education materials in particular, systematic non-compliers were not less likely to be exposed to printed material (OR 0.57, 95% CI 0.25–1.29), radio (OR 0.62, 95% CI 0.33–1.19), family members (OR 0.35, 95% CI 0.08–1.52), or LF staff (OR 0.50, 95% CI 0.23–1.08) than other persons.
| Sources of health education* | Percentage* | 95% CI |
|---|---|---|
| Person-to-person | 49.5 | 39.3–59.7 |
| Radio | 47.2 | 36.9–57.5 |
| Sound truck | 29.1 | 20.4–37.8 |
| Staff from hospital and LF program | 24.2 | 13.8–34.6 |
| Lymphatic filariasis film | 13.3 | 7.4–19.2 |
| Posters | 11.0 | 6.5–15.5 |
| Church | 9.8 | 5.0–14.6 |
| School | 6.9 | 4.2–9.6 |
| Banners | 6.9 | 3.2–10.6 |
| Family members | 6.6 | 4.0–9.2 |
| Health workers | 5.9 | 2.0–9.8 |
- * Respondents could mention several sources of information.
Rural–urban comparison: total population survey (MDA 3)
In the sample for the total population and adult survey, only five of the 14 urban distribution posts were randomly selected. To be able to conduct the rural–urban comparison, we added information collected from the nine remaining urban distribution posts. We compared information from 647 persons living in 14 urban clusters with 1530 persons living in 35 rural clusters. The median family size was five persons per family in urban settings (range 1–20) and six (range 1–12) in rural settings. There was no rural–urban difference in median age (12 years) or gender (54% female). The coverage in the urban area for the third MDA was 68.3% and in rural areas it was 78.3% (P < 0.05) (Figure 2). Men were more likely than women to take the drugs (OR 1.44 for both urban and rural areas) but this difference was only significant in rural areas.
Age related coverage estimates and 95% confidence intervals for urban and rural settings, total population survey, third mass drug administrations, Haiti, Leogane. Data represent between five and 299 persons per age group.
Rural–urban comparison: adult survey (MDA 1–3)
For the adult survey, 339 adults (>14 years of age) were interviewed in rural areas and 137 adults in the urban area. There was no difference in median age or gender balance compared with total population survey. The urban population was significantly more educated and more likely to possess a means of transportation than the rural population (Table 5). Persons living in the city took the drugs as frequently as people in rural areas (Table 6) and there was also no statistically significant difference in the proportion of persons who never took the drugs between urban (18.5%, 95% CI 12.0–25.1) and rural settings (20.6%, 95% CI 14.1–27.1). The reasons not to take the drugs were similar for both settings, with the exception that the people in the city were more likely to have been absent during the MDA compared with people in rural areas (OR 3.26, 95% CI 1.5–7.1).
| Urban (%) n = 137 | Rural (%) n = 339 | P-value | |
|---|---|---|---|
| Education (highest level attained) | |||
| None | 9.0 | 16.2 | <0.05 |
| Primary school | 26.1 | 44.8 | <0.05 |
| Secondary school | 61.9 | 38.5 | <0.05 |
| University | 3.0 | 0.6 | <0.05 |
| Means of transport in household | |||
| Bicycle | 79.1 | 62.6 | <0.05 |
| Motorbike | 7.6 | 4.5 | 0.46 |
| Car | 15.8 | 4.8 | <0.05 |
| Participated in MDA | Urban (%) n = 137 | Rural (%)*n = 339 |
|---|---|---|
| Only MDA 1 | 8.1 | 5.8 |
| Only MDA 2 | 2.9 | 1.8 |
| Only MDA 3 | 5.2 | 9.7 |
| Only MDA 1 and 2 | 11.8 | 6.7 |
| Only MDA 1 and 3 | 2.9 | 5.2 |
| Only MDA 2 and 3 | 8.1 | 10.9 |
| 1, 2 and 3 | 40.4 | 41.3 |
| Never | 20.6 | 18.5 |
- * No statistically significant difference between urban and rural areas, for any comparison.
Discussion
Lymphatic filariasis is one of the six diseases considered by the International Task Force for Disease Eradication as potentially eradicable (MMWR 1993). To interrupt filariasis transmission, yearly MDAs with DEC or ivermectin, combined with albendazole, are recommended (Ottesen et al. 1997). There is no broad consensus about what level of drug coverage is needed to interrupt transmission of LF. Modeling for LF suggested that this level was 65% of the eligible population, but these models assume only partially systematic non-compliance (Stolk et al. 2003).
We conducted the adult survey to quantify the degree of systematic non-compliance and to understand the reasons for it. We found that 18.6% of the adult inhabitants from the plains of Leogane Commune had never participated in any of the three MDAs. Persons who never participated in the MDAs were not concentrated in a single demographic group; they were not older, more highly educated or concentrated in urban areas, as we anticipated. Because participation in the MDAs was greater for children than for adults, as shown in Figure 1 and described for the same population by Mathieu et al., the true level of non-compliance is likely to be lower than 18% (Mathieu et al. 2003). We cannot be certain that systematic non-compliance is jeopardizing the elimination effort, because we do not have information about the infection status of these persons. However, it would be unlikely that persons which were systematically non-compliant would have lower infection levels than those who participated in the MDAs. Consequently non-compliant persons represent a potential reservoir of infection that may contribute to ongoing transmission. More research is needed to identify and characterise persons who never participate in an MDA.
Based on a previous study conducted in Leogane, we know that knowledge of LF is directly related to participation in a MDA (Mathieu et al. 2004). We indeed found that the persons who systematically do not comply were less informed about the fact that tablets could prevent LF and that the tablets must be taken yearly (Table 3). However, we did not find that they were less exposed to health education materials. This suggests that non-compliant persons were exposed to the health education but did not retain or accept the health messages. No significant difference in education level or SES was found that would facilitate specific targeted health education. This leaves us with many questions about the nature of non-compliance and the factors responsible for this behaviour. With respect to mass drug administration programs, little is known about the nature of systematic non-compliance. Biritwum noted that between 16% and 34% never received drugs during the two or three MDAs in the Onchocerciasis Control Program in Benin, Ghana, Ivory Coast and Togo, but did not give a specific explanation for this (Biritwum et al. 1997). Systematic non-compliance in a community-wide vaccination program is often because of opposition to vaccination due to religious beliefs or to rumors (Andre 2003), however we have no evidence to suggest that similar explanations apply to MDA for LF in Leogane.
In this survey, we also assessed the impact of adverse events on participation in subsequent MDAs. Based on informal reports and focus groups in Leogane, adverse events following the first MDA were thought to have negatively influenced the decision to take the drugs during the second MDA, leading to a significant drop in coverage from year 1 to 2. We did not find a significant association between the reporting of side effects by respondents and non-participation in the following MDA.
One of the limitations in our study was that five members of the survey team were health educators who were also part of the social mobilisation team of the project. This could have introduced a bias as some of the persons interviewed may have recognised the interviewers. This, and the fact that we collected self-reported compliance data, may have introduced bias in reported compliance. There was likely to be a recall problem in the collection of information about participation during an MDA 3 years ago, although the coverage estimate for persons > 14 years of age found in the KAP survey conducted after the first MDA (63.9, 95% CI 58.5–69.3) is not statistically different from the coverage for the same MDA in the same age group found in this survey (59.4%, 95% CI 52.0–66.7) (Mathieu et al. 2004). On the other hand, there are no other large, highly publicised MDA programs in Leogane, so people would tend to remember each MDA. Although the reported coverage was much lower the second year, we did not find this to be the case for surveyed coverage. Except for problems of recall, no other likely explanation can be given for this finding. The difference in coverage among persons >14 years of age based on the general survey and the adult survey are likely due to a difference in gender representation in the two surveys.
Another limitation is that the study did not allow us to assess the impact of non-compliant persons on W. bancrofti transmission in part because we do not know their infection status. In addition, we cannot conclude that our estimate of 18% non-compliance in adults also applies to children because we excluded persons from 14 years of age or less from the adult survey. However, given that coverage of school aged children was higher than in the population overall and that children make up a substantial proportion of the total population, the overall rate of non-compliance is likely to be <18%.
Further research is needed to characterise the non-compliant persons and their impact on LF elimination programs. Additional efforts are also needed to understand how to address non-compliance with new health education and drug distribution strategies.
Acknowledgements
We thank Dr J. Lafontant, director Hôpital St Croix, Leogane, Haiti, the survey team, and all other staff members of the filariasis elimination project, Leogane, Haiti for their commitment and hard work. Funding for this project was provided by CDC's Emerging Infection Program and by a grant from the Bill and Melinda Gates Foundation to the University of Notre Dame.
