| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
 |
ABSTRACT |
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES |
|---|
|
INTRODUCTION |
|
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES |
|---|
Malaria control in Haiti has primarily relied on treatment of confirmed cases with chloroquine, although mass chloroquine treatment and indoor-residual spraying (IRS) aimed at eradication were both attempted with limited success in the 1960s–1980s.4,5 It has been surmised that these campaigns were largely unsuccessful because of a general lack of data on malaria transmission to guide limited resources towards the most affected areas.6,7 More recently, in 2005, Haiti was awarded funding for malaria control from the Global Fund to Fight AIDS, Tuberculosis, and Malaria (GFATM), with efforts expected to include free and subsidized distribution of insecticide-treated bed nets (ITNs), enhanced surveillance, and case management. Although larval control is currently being done on a limited basis, additional vector control activities are planned.8 However, there remains a paucity of reliable epidemiologic data on the distribution, burden, and factors related to malaria within Haiti to guide planned control efforts.6
Numerous studies have been conducted globally that describe socioeconomic, sociodemographic, and environmental risk factors associated with malaria and malaria-related knowledge, perceptions, and prevention practices.9–18 Seasonality, irrigation, farming, poverty, education, house construction, distance to water bodies, and the location of residences have been linked to malaria transmission; proximity to health clinics, perception of risk, assets, and education have been shown to be important determinants of ITN use and personal protection practices against malaria. Net use and personal protection have also been linked to simple avoidance of mosquitoes, irrespective of knowledge of malaria risk.19 Evidence from Haiti suggests that age, sex, the composition of the household, and the expectations associated with disease prognosis influence malaria control through conceptual understandings of causes of fever and convulsions.20 Some regional studies in central America also show the importance of understanding how sociodemographic factors such as poverty, education, age, history of fever, and knowledge of malaria transmission effect operational aspects and sustainability of malaria control.21,22 Understanding who already knows about malaria and malaria prevention, who has adopted malaria prevention and mosquito avoidance practices, and who is at risk of malaria infection is a necessary precursor to identifying and targeting vulnerable populations and ensuring successful implementation and sustainability of malaria control efforts.23–25
We explored the Haitian context to identify factors that may influence the successful implementation of malaria control in the country. The specific objectives are to 1) describe malaria-related knowledge, perceptions, and practices, and 2) determine which sociodemographic factors are associated with malaria infection. Ethical approval for this study was provided by the Tulane University Institutional Review Board (IRB) and Hôpital Albert Schweitzer (HAS).
|
METHODS |
|
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES |
|---|
).2,4,5 This low-lying area is heavily irrigated from surrounding mountains and the Artibonite River system, with rice production being one of the most common forms of agriculture. Most houses are two-room structures made of concrete with tin or thatched roofs. The primary malaria parasite transmission season is November through January.4 The prevalence of P. falciparum malaria infection was estimated to be 3.1% in November–December 2006.26
|
Study design and sample size. A sample size of 800 persons for estimating malaria parasite infections was sought based on the desire of a maximum tolerable error of ± 3%, a design effect of 2, probability of committing a type-1 error = 5% (two-tailed), an estimated population prevalence of 10%, and a non-response of 10%. Information on malaria-related knowledge, perceptions, and practices was ascertained from the head of 200 households on the basis of the following parameters: a maximum tolerable error of ± 10%, a design effect of 2, probability of committing a type-error = 5% (two-tailed), an estimated population proportion of 50% and a non-response of 10%.
A two-stage cluster design was used to generate a probability sample of households within the study area of the Artibonite Valley during the high malaria transmission season in November–December 2006. Because of logistical constraints and limited access, all villages at altitudes higher than 500 meters above sea level were excluded from the sample frame. Twenty primary sampling units (PSUs) were selected in the first stage and 10 households selected from each PSU at the second stage, totaling 200 houses. The sampling frame consisted of a detailed list of all villages in the Artibonite Valley served by HAS, which included the approximate numbers of households within each village established through the HAS community surveillance system. At the first stage, 20 PSUs were selected based on probability proportional to the relative number of households within each PSU. Prior to second stage sampling, all households within each selected PSU were enumerated by the HAS data collection team to create the second-stage sampling frame. A total of 200 respondents, one per selected household, constituted the household sample; 714 persons from within these 200 households provided blood for parasite diagnosis.
Parasitologic and clinical data. Thick and thin blood slides, four blots of blood on filter paper, and axillary temperature were obtained from all consenting persons more than one month of age in each selected household; informed consent was obtained from all persons or caregivers of persons less than 15 years of age before they provided blood. All 714 blood slides were examined by microscopy by a trained laboratory technician using standard methods.11 Polymerase chain reaction (PCR) was then used to test for P. falciparum parasites at Tulane University laboratories, as described elsewhere.26 Persons positive for malaria parasites by microscopy or PCR were treated with chloroquine, which is the standard treatment for malaria in Haiti.
Household demographic data. Household demographic data were collected from an available adult more than 15 years of age residing in the household using a standardized questionnaire. Data included a complete listing of children and adults who slept in the house the previous night, ownership and use of bed nets and ITNs, use of other malaria control methods, history of fever, treatment-seeking behavior, knowledge of the route of malaria parasite transmission, mosquito control practices, an inventory of household possessions for calculating a household asset index, and basic demographic data such as age, sex, and education of the head of household. Mothers of children less than five years of age were then interviewed to ascertain treatment-seeking behaviors for fevers in their children in the past two weeks.
Normalized difference vegetation index and elevation data. Elevation data for all selected villages were obtained from the U.S. Geological Survey through their Seamless Data Distribution Interface International Data link and merged with village latitude and longitude data using ArcGIS software® (Environmental Sciences Research Institute, Redlands, CA). The normalized difference vegetation index (NDVI) was calculated from Landsat 7–Enhanced Thematic Mapper (ETM) data sets obtained from the U.S. Geological Survey for August–December 2006. ETM data were collected at 28.5-meter spatial resolution. Bands 3 (red) and 4 (infrared) of the ETM were used in the calculation. The calculation used was NDVI = (([Band 4] – [Band 3])/([Band 4] + [Band 3]).Float). This formula produced a floating point format ranging from -1 to 1. Values approaching -1 represent the absence of photosynthesizing material such as clear water or built infrastructure; values approaching 1 represent the presence of large amounts of photosynthesizing material such as lush, vegetated areas.
Data analysis. All analyses were done using STATA version 9.0 (Stata Corporation, College Station, TX). Descriptive statistics were used to summarize household survey data by select demographic characteristics. Chi-square and Fisher’s exact test statistics were used to assess differences in bivariate outcomes. A household asset index was developed based on a principal components analysis of household assets.27 Reported household assets included a refrigerator, access to electricity, radio, television, running water, bicycle, car, motorcycle, an indoor toilet, and a telephone. The raw factor scores were then categorized at the median value into lower and higher asset categories.
To control for clustering, the Huber-White-Sandwich estimator of variance was used to calculate empirically estimated standard errors for all population point estimates; generalized estimating equations were used to obtain empirically estimated standard errors within logistic regression models, using the village as the cluster unit. All point estimates were weighted to account for discrepancies in estimated versus actual village sizes at first stage selection. Wald statistics and log-likelihood ratios were used to identify variable significance and model fit with the probability of committing a type-1 error (
) set at 0.05.
Logistic regression was used to test whether sociodemographic factors and the presence of a case of malaria parasite infection in the household were associated with the outcome of correct knowledge of malaria parasite transmission, risk, and prevention (n = 200). Correct knowledge was defined by the household respondent reporting all three of the following: malaria parasites are transmitted by mosquitoes, both children and adults are at risk, and ITNs can be used to prevent malaria parasite infection. On the basis of previous research,23–25 independent variables included in this model were age, sex, education, household asset, and the presence of a case of malaria infection in the household. Village level control variables included the mean NDVI value, a proximate determinant of the propensity of an area to harbor mosquitoes and village elevation. Given that the roads leading up into the mountains are in poor condition with limited public transportation services, few health facilities exist at higher elevations, and many residents report receiving health and malaria information from hospitals and clinics, village level elevation is used here as a proximate determinant of access to healthcare and malaria prevention information (HAS community health workers, unpublished data).
A second logistic regression was performed to assess risk factors of P. falciparum infection identified by PCR (n = 714). On the basis of previous research,23–25 independent variables included in this analysis were age, sex, presence of fever, head of household education, and household asset index, and village level control variables included mean NDVI and village elevation.
|
RESULTS |
|
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES |
|---|
A total of 200 household respondents were included in the sample: 74 men and 126 women. Of the 804 persons more than one month of age within the randomly selected households, 714 (age range = 1–92 years) provided a blood sample for malaria parasite diagnosis (329 men and 385 women) (Table 1). There was no household non-response; individual non-response was 11.2%. Individual non-response was the result of refusal to provide a blood sample for malaria parasite diagnosis.
|
Seven (< 1%) P. falciparum infections were identified by microscopy. Twenty-three P. falciparum infections were identified by PCR (3.1%),26 with all cases concentrated in 8 (40.0%) of the selected villages. A total of 16 households had at least 1 resident test positive for malaria parasite infection. All seven infections identified by microscopy were also positive by PCR. This discrepancy suggests low sensitivity of microscopy for detecting malaria parasites in this context.
Knowledge of malaria transmission.
Most respondents thought that malaria was a problem in their community (78.1%) and correctly cited that mosquitoes are responsible for transmission (68.1%) (Table 2). However, nearly half (47.0%) also incorrectly cited that contaminated water was a major cause, and 14.5% reported that unhygienic surroundings caused malaria. Respondents from households with more assets (
2 = 16.01, P < 0.001) and with more education (2 = 4.81, P < 0.05) were significantly more likely to correctly cite that mosquitoes cause malaria. Nearly all respondents (88.9%) knew that children and adults are at risk for malaria.
|
). Respondents more than 60 years of age were much less likely to have correct knowledge related to malaria than respondents 
60 years of age (adjusted odds ratio [OR] = 0.24, 95% confidence interval [CI] = 0.07–0.81). Respondents residing in households with more assets were significantly more likely than those with fewer assets to have correct knowledge about malaria (OR = 8.3, 95% CI = 3.30–20.64). Although marginally insignificant, respondents residing in a household where a malaria infection was detected were less likely than those in houses with no malaria infection to have correct knowledge of malaria (OR = 0.20, 95% CI = 0.03–1.13).
|
). Accordingly, net and ITN use was low, with only 3.6% and 2.7% of children less than five years of age reported to have slept under a bed net and ITN, respectively, the night before the survey.
|
2 = 1.6, P = 0.20). However, less than 9% reported willingness to pay more than 150 Haitian Gourdes (approximately U.S. $4.00): most reported a willingness to pay between 25 and 100 Haitian Gourdes (approximately U.S. $0.65 and $2.63). One (25.0%) of four respondents reported knowledge of at least one place to obtain an ITN; no significant association was detected between knowing where to obtain an ITN and household assets or education sub-groups. However, a significant difference was detected between male (31%) and female (15%) respondents reporting to know where to obtain an ITN (2 = 6.6, P < 0.01). Less than one-fourth (21.7%) of the respondents knew that ITNs reduced the risk of malaria parasite transmission. Information on ITNs was most commonly obtained at the clinic or hospital (52.0%) and from community health workers (41.0%). One-fourth (27.0%) reported hearing at least one message about ITNs on the radio in the last six months, and hardly anyone (< 1.0%) reported hearing or reading ITN messages on the television or in the newspaper.
Few respondents (3%) reported mosquitoes to be a nuisance where they live. Accordingly, personal efforts toward malaria control were infrequent with less than 2% reporting taking action to stop mosquitoes from biting, and only 1% reported removing standing water in the community to combat mosquitoes over the last 12 months.
Of the 98 children less than 5 years of age, two-thirds (65.0%) had a fever or convulsions in the two weeks preceding the survey (Table 5). Presumptive treatment of fevers with an anti-malarial drug was low, with only 9.0% given an anti-malarial drug; 2.1% took an anti-malarial drug the same or next day after the onset of the fever. All treatment was sought at a hospital or clinic. Similar to ITNs, the most common medium for getting information about anti-malarial drugs came from community health workers (43.0%) and clinics and hospitals (32.0%). Seventeen percent of respondents reported hearing at least one message about anti-malarial treatments on the radio in the last six months, and few reported hearing such messages on television (1.0%) or through the newspaper (1.7%).
|
|
DISCUSSION |
|
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES |
|---|
Although most respondents knew that mosquitoes are responsible for transmitting malaria, and most thought malaria was a problem in their community, ITN ownership and use was low, which is similar to findings in Haiti observed previously.28 This finding is not surprising because only one in five respondents knew that ITNs reduce the risk of malaria parasite transmission, only one in four knew where to get an ITN, and few respondents (3%) reported mosquitoes to be a nuisance where they live. Although most respondents said they would be willing to pay for an ITN, few said they would be willing to pay more than US $4, and most said they would not be willing to pay more than US $2.63 (100 Haitian Gourdes). The nationally set sales price for an ITN is 200 Haitian Gourdes, well above the thresholds cited by the households. Such data suggest that ITNs will need to be heavily subsidized to increase coverage. As ITN distribution is scaled up in Haiti, these data also suggest information campaigns should focus on getting the word out that ITNs are effective for reducing the risk of encountering an infective mosquito, in addition to nuisance-biting mosquitoes. Emphasis should also be placed on disseminating information about where an ITN can be obtained, coupled with strategically placing distribution centers in areas with high visibility.
This study found that respondents from households where a P. falciparum infection was found were less likely to have correct malaria-related knowledge. This information is useful because it provides a starting point for identifying persons who need information about malaria parasite transmission and treatment. Identifying the communities where persons positive for malaria infections reside could provide insight into which areas have suitable ecologic conditions for mosquito proliferation, which would be useful for targeting areas for IRS and/or ITNs. The strong association between assets and correct knowledge of malaria parasite transmission suggests traditional mass-media strategies for disseminating health information may not be reaching the poorest populations, where TV and radio ownership, and newspaper readership is low. As well, most respondents reported getting health and malaria information from community health workers and staff at hospitals or clinics. Thus, expanding the reach and scope of the current health system information dissemination strategies to include additional clinics or communities may be highly successful in increasing malaria-related knowledge. The lack of a significant association between education and correct knowledge further suggests the need to develop non-traditional channels for disseminating newer health-related information.
Presumptive home treatment of fevers with chloroquine is not the recommended policy in Haiti because transmission is low with most fevers unrelated to malaria. A policy of presumptive home treatment of malaria is not advocated here. However, results show a strong association between having a P. falciparum infection and being symptomatic with fever. Results also show that among children receiving malaria treatment of fevers, all received treatment at a hospital or clinic, although few received treatment within one or two days of the onset of fever. These factors are important because P. falciparum is still chloroquine sensitive in Haiti; thus, strong surveillance and case management will be effective as long as persons seek prompt treatment of their fevers. Although barriers to accessing healthcare and treatment-seeking behaviors persist throughout Haiti, there may be potential for interventions that focus on recognition of symptoms and the importance of seeking prompt diagnosis and treatment because most of the population is immunologically naive to malaria. Thus, signs and symptoms are most often distinct and severe.
The following factors should be considered prior to scaling-up ITN distribution or IRS programs in low-transmission settings such as Haiti. First, previous studies have shown An. albimanus to be primarily exophilic, preferring to rest outdoors.3,29,30 Thus, the effectiveness of ITNs and IRS may be limited in the absence of novel and highly focal ITN distribution or IRS application strategies. Second, there have been few ITN efficacy trials to this point in areas of unstable P. falciparum malaria transmission such as Haiti.31 Trials assessing the impact of IRS have been conducted in areas of mostly stable malaria transmission in sub-Saharan Africa where the vector is known to be endophilic.32–36 Third, the observed lack of knowledge that ITNs are an effective means of preventing malaria along with the reluctance to pay for one suggests scaling up ITN distribution among the general population poses considerable challenges. And lastly, although ITNs and IRS have been shown to be cost-effective strategies for controlling malaria in areas of stable transmission,37,38 few studies have documented their cost-effectiveness in areas of low transmission, especially compared with surveillance and case management in areas with chloroquine sensitivity such as Haiti. It is likely that as the incidence of malaria decreases significantly, as one would observe when comparing high-transmission settings to low-transmission settings, the cost associated with each case averted would increase considerably because of the decreasing marginal return of preventative strategies such as ITNs and IRS.38
As with most epidemiologic field studies, several methodologic limitations are worth noting. First, results are generalizable only to villages within the sampling frame of the Artibonite Valley, which excluded villages at altitudes higher than 500 meters. Second, only data on health-seeking behavior for children less than five years of age were collected. Thus, fever and health-seeking pattern results are not generalizable to older persons. Third, use of a household asset index to capture the relative wealth of the household ignores the fact that poor persons can live in wealthier households, and multiple families or extended family may also live in the house. However, the asset index is subject to less recall bias and less misclassification of relative wealth than the measurement of income or consumption patterns within households. Last, the power of this study was set to estimate malaria parasite point prevalence26; given the low number of infections found, a larger sample size may be needed in low-transmission settings to strengthen the external validity of individual-level risk factor analyses.
In conclusion, we argue that malaria control in Haiti should focus on enhanced surveillance and case management, coupled with expanded information campaigns about malaria prevention options and treatment of fevers, and that ITNs and IRS programs should be highly targeted to those communities most at risk in Haiti. We propose that enhanced surveillance systems at hospitals and clinics could be used to identify persons, households, and villages at the greatest risk of malaria. Thus, these areas and persons should be given priority malaria prevention interventions.
Received June 27, 2007. Accepted for publication November 10, 2007.
Acknowledgments: We thank the communities that participated in this study for their cooperation; the data collectors, scientists, administration, and laboratory staff at Hôpital Albert Schweitzer for their collaboration and continued support of public health research; Don Krogstad for his advice, supervision, and use of his laboratory at the Department of Tropical Medicine at Tulane University for the PCR work; Chris Swalm (Tulane University) for his assistance in geo-processing the NDVI and elevation data; two anonymous reviewers whose comments greatly improved this manuscript; Olbeg Desinor (United States Agency for International Development/Haiti) for his support of this work; and the Haitian Ministry of Health for allowing us to conduct this research in Haiti.
Financial support: This study was supported in part by the United States Agency for International Development through a subcontract with Research Triangle Institute, and the Tulane University Research Enhancement Fund.
Disclaimer: The opinions and assertions expressed herein are those of the authors and do not necessarily reflect the official position or policy of the United States Agency for International Development, Tulane University, or Hôpital Albert Schweitzer.
* Address correspondence to Joseph Keating, Department of International Health and Development, Tulane University School of Public Health and Tropical Medicine, 1440 Canal Street, Suite 2200, New Orleans, LA 70112. E-mail: jkeating{at}tulane.edu 
Authors’ addresses: Joseph Keating, Thomas P. Eisele, Adam Bennett, and Kate Macintyre, Department of International Health and Development, Tulane University School of Public Health and Tropical Medicine, 1440 Canal Street, Suite 2200, New Orleans, LA, 70112, Telephone: 504-988-3655, Fax: 504-988-3653. Dawn Johnson, Community Development Division, Hôpital Albert Schweitzer, Deschapelles, Haiti.
Reprint requests: Joseph Keating, Department of International Health and Development, Tulane University School of Public Health and Tropical Medicine, 1440 Canal Street, Suite 2200, New Orleans, LA, 70112, Telephone: 504-988-1458, Fax: 504-988-3653, E-mail: jkeating{at}tulane.edu.
|
REFERENCES |
|
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES |
|---|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
