COMMUNITY-BASED DENGUE PREVENTION PROGRAMS IN PUERTO RICO:

(1)

COMMUNITY-BASED DENGUE PREVENTION PROGRAMS IN PUERTO RICO:

IMPACT ON KNOWLEDGE, BEHAVIOR, AND RESIDENTIAL MOSQUITO INFESTATION

PETER J. WINCH, ELLI LEONTSINI, JOSE´ G. RIGAU-PE´REZ, MERVIN RUIZ-PE´REZ, GARY G. CLARK,AND

DUANE J. GUBLER

Department of International Health, School of Hygiene and Public Health, The Johns Hopkins University, Baltimore, Maryland;

Dengue Branch, Division of Vector-Borne Infectious Diseases, Centers for Disease Control and Prevention, San Juan, Puerto Rico;

Division of Vector-Borne Infectious Diseases, Centers for Disease Control and Prevention, Fort Collins, Colorado

Abstract. Dengue is a major health burdeninPuerto Rico. Televised public service announcements and posters, elementary and pre-school educational programs, and an exhibit at the Children’s Museum inOld SanJuanwere evaluated separately using knowledge and practices surveys administered to children and their parents, surveys of house lots for larval container habitats, focus groups, and interviews with program organizers and participants. Exposure to the programs was associated with increased dengue-related knowledge, increased proportionof tires protected from rain, decreased proportion of water storage containers positive for mosquito larvae, and increased indoor use of aerosol insecticides. Exposure to the elementary school program was associated with slightly lower indices of residential mosquito infestation. The programs have resulted in high levels of awareness, some behavior change, and limited change in larval indices. Greater emphasis on the skills necessary for community members to keep containers free of mosquito larvae would increase program effectiveness.

INTRODUCTION

Dengue is an acute viral disease that causes fever, severe headache, bone and joint pains, nausea, vomiting, and pros- tration. In some cases, it may produce hemorrhagic manifes- tations, shock, and death.¹The principal vector is the mos- quito Aedes aegypti, which has a worldwide distributionin tropical and many semitropical areas. The modern era of dengue control inPuerto Rico beganin1963, whenanepidemic of dengue-3 virus resulted in 27,000 reported cases.² Radio announcements and slides for use in movie theaters were developed to educate and promote public involvement in mosquito control, and a survey documented high levels of awareness, with 72% of 799 respondents having heard or read something about dengue, and 65% stating that it was caused by a mosquito.³Despite the public’s awareness, the disease became endemic, with frequent epidemics after 1977. Single cases of dengue hemorrhagic fever (DHF) were documented in1975, 1982, and in1985.^4,5 Inresponse to the threat of DHF, the Dengue Branch of the Centers for Disease Control and Prevention (CDC) and the Puerto Rico Department of Health developed an integrated approach to dengue control in the early 1980s that encompassed 1) entomologic and laboratory-based disease surveillance; 2) contingency plans for emergency mosquito control and hospitalization of large numbers of people during epidemics; 3) education of the medical community; and 4) pilot community-based dengue prevention programs promoting the elimination or control by the public of containers found in and around the home such as tires, metal cans, and house plants that harbor larvae of the Ae. aegypti mosquito.⁶

Inthe late 1980s and 1990s, there was anincrease inthe number of DHF cases reported. In the largest recent epidemic (1994–1995), 23,673 cases of dengue (6.7 per 1,000 population) were reported to the laboratory-based surveillance system.⁷While mortality is low (annual average⳱ four laboratory-positive deaths, 1991–1995),⁸ the burdenof morbidity attributable to dengue transmission in Puerto Rico is substan- tial. From 1984 to 1994, dengue produced a loss of disability- adjusted life years equivalent to the burden imposed (per

millionpopulation) by malaria or tuberculosis inLatin Americancountries.⁸

The increases in dengue incidence in Puerto Rico and throughout Latin America prompted the need to determine the impact of the pilot community educationprograms inthe target populations. This paper will present the methods and some of the results of a comprehensive cross-sectional evaluation of the pilot community-based dengue prevention programs inPuerto Rico conducted in1995. The principal objec- tives of the evaluationwere to document the process by which the program has been implemented by different organizations; to measure current levels of knowledge of dengue transmission, prevention, and treatment among program participants; and to assess the extent to which knowledge about dengue is reflected in the adequate control of mosquito larval habitats.

MATERIALS AND METHODS

The Caribbeanisland of Puerto Rico (18°15⬘N, 66° 30⬘W) is approximately 110 miles long and 35 miles wide, with a moun- tainous central area, an average annual temperature of 25°C, average total annual precipitation of 1,780 mm, and a population density of 396 persons/km².⁹

Program components. Four components of the community- based prevention programs were assessed. The first was a Head Start module initiated in 1986. Head Start is a federally funded program to improve the health and development of low income preschool children. The children’s component consists of classroom and community activities (games, drawings, parades) described inanactivity book, which highlights mosquitoes, their breeding habitats, and their potential dan- ger to public health. The parents’ component consists of vis- iting the Head Start center for a lecture and/or video on dengue prevention as motivation to search for and control larval habitats ontheir property, and home visits by Head Start personnel.

The second component was a public school program initiated in1987. As part of the regular fourth grade social science 363

(2)

curriculum, students become involved in exercises, inside and outside the classroom, outlined in a dengue pamphlet. The primary channel for reaching parents is the child, who is en- couraged to share with them what was learned in school about dengue. Part of the program requires that the student look for actual or potential Ae. aegypti larval habitats at home and destroy or control them while pointing them out to the parents.

The third component was posters and televised Public Ser- vice Announcements (PSAs). These have been developed and disseminated in collaboration with members of the Ro- tary Club of SanJuansince 1985. The latest poster, “Dengue- Free Zone,” which was widely distributed in schools, phar- macies, and at community events, stresses safe water storage as one of the four behaviors depicted (Figure 1).

The fourth component was the Children’s Museum exhibit on Ae. aegypti. This is the most recent community-based pro- gram, established in1993 whenthis new Museum opened its doors to the public inOld SanJuan. It was estimated by Museum staff that 40,000 childrenvisited the Museum in 1995. The exhibit was constructed on the building’s roof and resembled a little house with its patio. Glass cases with live Ae. aegypti immature and adult mosquitoes were displayed, and commonly found larval habitats infested with larvae were staged inthe patio. The young visitors begantheir visit by looking at the glass cases and learning from a museum facilitator about the mosquito life stages and dengue transmission.

They would then enter the little house and watch a seven- minute video, which modeled source reduction behaviors by kids, such as emptying water from buckets, drinking dishes for dogs, or flowerpots. The facilitator would thenwalk the visitors through the larval habitats inthe patio, point out the larvae swimming in them, and hand out droppers to the childrenand show them how to collect larvae out of the water to be able to see them better.

Selection of treated and untreated groups. With the excep- tion of the Museum exhibit, no baseline data were available to measure program impact through a comparison of conditions before and after the implementation of the pilot community- based programs. Therefore, surveys included questions to assess exposure to different sources of information, and the analyses used degree of exposure to the different sources as independent variables, in addition to the dichotomous variable for treated/untreated group. Public Service Announce- ments were not analyzed as a distinct intervention component because it was assumed that individuals from both the treated and untreated groups would have been exposed to them.

For each dengue educational module (Head Start and elementary school), a treated or exposed site was identified where the module had beenfully implemented, as well as community-wide events, where such events as anti-dengue days, health festivals, and clean-up campaigns, had been held.

The untreated groups were selected based on the following criteria: 1) No implementation of the program component being evaluated, 2) Similar distribution of rural and urban housing types, 3) Similar topography and rainfall, and 4) Simi- lar socioeconomic status of the population. Participants in both programs had beenexposed to the module 3–12 months before responding to the survey. Sites were selected in a large enough group of Head Start centers or school to yield a sample size of at least 100 childrenand their respective parents for the survey measuring knowledge and exposure, and

as close to 250 households as possible for the larval infestation survey, to allow for the detection of a minimum difference between the treated and untreated group of 13–15%. Each treated site was paired with an untreated one that fulfilled these sample size requirements and was comparable in social and economic variables (1990 census data), in access to basic services such as potable water, and was ecologically similar (elevation, rainfall, vegetation, and rural or urban place- ment).

For evaluationof the elementary school program, the treated group was identified in the adjacent municipalities of Moca (population⳱ 33,000)⁹and Aguadilla (population⳱ 59,000)⁹ inwesternPuerto Rico. The untreated group was identified in the municipality of Cidra (population⳱ 36,000), southwest of San Juan. Municipalities in Puerto Rico are equivalent to counties in the United States, encompassing both an urban center and rural areas, and Moca, Aguadilla, and Cidra all have a mixture of rural and urban housing types.

For evaluationof the Head Start dengue program, the treated group was identified in the municipality of Carolina (popula- tion⳱ 178,000),⁹located onthe east side of metropolitanSan Juan. The untreated group was identified in the municipality of Caguas (population⳱ 133,000),⁹south of SanJuan. Both municipalities had predominantly urban housing types.

For evaluation of the Museum exhibit, entry and exit polls were administered to two, initially equal, random samples of entering and exiting visitors, respectively, 5–15 years old, vis- iting the Children’s Museum between July 19 and August 2, 1995.

Data collection. Informed consent was obtained from all adult participants and from parents or legal guardians of mi- nors. The evaluation protocol, all consent and assent forms, and informational materials were approved by the CDC’s Of- fice of Program Planning and Evaluation, Office of Program Resources, and Privacy Act Coordinator’s Office; the Johns Hopkins University Office of Human Research; and the Fed- eral Office of Management and Budget (approval no. 0920- 0360, February 21, 1995).

Three methods of data collectionwere used. The first was qualitative research, in which semi-structured interviews and focus groups were conducted with parents, teachers, and of- ficials from Head Start programs and schools exposed to the dengue module and Museum exhibit facilitators. The second method was surveys on knowledge and exposure, incorporat- ing results of the qualitative research. These were administered inthe classroom to childrenintreated and untreated Head Start centers and public schools (grades 4 and 5), to their parents at home, and at the Children’s Museum amphi- theater or courtyard for the Ae. aegypti exhibit. Four instru- ments were administered, one to elementary school students, another to Head Start children, a single instrument to all parents, and the same instrument to both entering and exiting Museum visitors, with additional exposure questions for the exit poll. Photographs (5⬙ × 7⬙ prints) accompanied many questions to standardize clarity; for example, whether different water collection sites are potential sources of mosquitoes.

Vials containing preserved larvae were presented to all respondents for identification. All interview instruments collected general information such as age, educational level, and place of residence, and assessed exposure to specific program activities, exposure to information about dengue prevention from other sources (e.g., physicians, advertisements on buses,

(3)

FIGURE1. A recent “Dengue-Free Zone” poster/flier used by the community-based dengue prevention programs in Puerto Rico.

(4)

Children’s Museum), and correct and incorrect knowledge about larval habitats, dengue transmission, and preventive measures. Giventhe age (4–5 years old) of the Head Start children, their instrument was the simplest and briefest to administer, and available choices were illustrated by drawings adapted or inspired from the Head Start program’s activity book. The parents’ survey also included questions on their awareness of their children’s program activities and commu- nicationwith their childrenabout dengue. The third method was larval surveys conducted in the homes of the surveyed children(except for Museum visitors), to enumerate all actual or potential container habitats for Ae. aegypti, and examine them for the presence of water and/or larvae. The following container types were defined as disposable: tin cans, small bottles, and small plastic containers used for food. Samples of the larvae were taken for species identification. The larval survey was administered at the same time as the knowledge and exposure survey (from April to June 1995). To adjust for seasonal differences in temperature and rainfall affecting mosquito infestation and the increase in the larval indices that occurred as the rainy season progressed, larval surveys ro- tated between exposed and unexposed municipalities, and the number of days since the beginning of the evaluation (1–89) was modeled as a predictor variable. This predictor variable was then entered first in multivariate analysis examining the effect of the various intervention components to further adjust for the effect of the timing of the survey on the probabil- ity of containers having water and/or mosquito larvae in them.

Outcomes. Following bivariate analysis (by treatment sta- tus) of all individual responses to knowledge and exposure surveys, multivariate analysis was conducted by constructing

seven additive knowledge indices and three additive exposure indices for parents (Table 1). The content of all indices was confirmed through reliability analysis.¹⁰ The transformed knowledge data were then analyzed by treatment status and, where applicable, by degree of the parent’s exposure to communication with child, posters or PSAs, controlling for con- founders such as age, gender, school grade, years of education, and employment status.

We examined the impact of the programs on overall Ae.

aegypti infestationlevels, as well as onfour specific behaviors:

elimination of refuse, protecting used tires from the rain (and thus from oviposition by mosquitoes), maintaining water storage containers free of larvae, and use of commercial indoor aerosol insecticides. Behavioral outcome variables were two widely used Ae. aegypti infestation indices: the House Index (percent of inspected premises found to contain larvae) and the Breteau Index (the mean number of containers found to contain larvae per premise, expressed as a percentage) as well as container-specific indices, such as total and mean number of a type of container, number protected, and number positive for any mosquito larva. Tires were not counted with the disposable containers because residents consider them to be useful. A protected tire was defined as one without water and found under a roof or cover. To measure correct water storage practices, we examined the percentage of drums, buckets, and bottles positive for any type of mosquito larva because the behaviors that maintain a water storage container free of Ae. aegypti (such as covering the container with a tight-fitting lid or emptying the container out completely before filling it againwith water) should also preclude the presence of larvae of other mosquito species.

TABLE1

Dengue-related knowledge and parents’ degree of exposure to the dengue programs*

Additive index description

Number of items inindex

Reliability (Chronbach’s

alpha)

Maximum possible

score

Comparisonof meanscores by treatment status

Treated group Untreated group

Statistical procedure

Elementary school children Moca-Aguadilla (n⳱ 162) Cidra (n⳱ 50) ANOVA

Correct overall knowledge 27 0.74 39 22.2 17.8 P < 0.001

Head Start (pre-school) childrenCarolina (n⳱ 128) Caguas (n⳱ 144)

Young visitors of Children’s Museum Exit poll (n⳱ 98) Entry poll (n⳱ 113) Multiple linear

exhibit on Aedes aegypti regression

Elementary school program Heart Start program

Treated Untreated Treated Untreated

Parents of elementary school and Head Start children (n⳱ 140) (n⳱ 132) (n⳱ 96) (n ⳱ 114)

Correct knowledge of preventive 5 0.77 5 4.62 4.64 4.71 4.62 Multiple linear

measures NS NS regression

Correct knowledge of larval habitats 8 0.64 16 12.85 12.87 12.82 12.70

NS NS

Incorrect knowledge of cause 5 0.79 15 8.30 7.44 7.19 6.77

of dengue P⳱ 0.025 NS

Incorrect knowledge of mosquito 8 0.62 11 4.92 5.92 5.78 6.07

life cycle P⳱ 0.002 NS

Communication with child 4 0.81 8 4.55 2.83 2.43 1.32

regarding dengue prevention P < 0.0001 P < 0.0001

Exposure to posters about 4 0.65 8 4.31 3.68 3.82 3.90

dengue prevention P⳱ 0.009 P⳱ NS

Exposure to Public Service 5 0.53 9 5.27 4.90 5.53 5.13

Announcements NS NS

* ANOVA⳱ analysis of variance; NS ⳱ not significant.

(5)

Behavioral outcome variables were compared by treatment status and by the parent’s degree of exposure to the programs, controlling for confounding by 1) seasonal changes in temperature and rainfall occurring as the larval surveys were completed, (summarized by the 89-day time variable), and 2) rural versus urban housing and patio characteristics that af- fect the type and number of potential larval habitats in a home (summarized by a housing type, a house lot size and a patio size variables). The standard deviations for the mean number of containers per premise were very high due to a few households having enormous numbers of containers. There- fore, the total and disposable container variables were log transformed.

RESULTS

Impact on knowledge. Exposure of childrento the elemen- tary school and Head Start programs, as well as the Children’s Museum exhibit, was associated with significantly higher levels of correct overall dengue-related knowledge (Table 1). In addition, the elementary school program and the Museum exhibit were associated with significantly lower incorrect children’s knowledge concerning the mosquito life cycle, disease transmission, larval habitats, and control measures. Children in both classroom-based programs were significantly more likely to discuss dengue with their parents thanchildreninthe untreated group, and, as expected due to age differences, levels of communication attained in elementary school were greater thanthose inHead Start (Table 1). These findings were supported by statements made in focus groups and interviews:

“. . . the teacher had given [my children] this assignment that they have to come and check the back yard, so they threw downtheir school bags and set off to check the patio; they started to throw buckets around; they started to check the dishes of the animals where you put their water. . . . [an d they said] ‘Mommy you can’t have any- thing because later the mosquito bites us and can make us sick’.”

Among parents, there was a high level of correct knowl- edge about dengue preventive measures and Ae. aegypti lar- val habitats, independent of treatment status (Table 1), as well as some incorrect knowledge. The child’s exposure to the elementary school program was associated with the parent’s lower incorrect knowledge of the mosquito life cycle, but paradoxically, higher incorrect knowledge of the cause of dengue (Table 1). Parents of exposed elementary school students more often accepted that sharing a glass with someone sick with dengue (38.6% in the treated group versus 27.3% in the untreated group; P⳱ 0.038, by chi-square test for linear trend) or sneezing by someone sick with dengue (45.7% ver- sus 29.5%, respectively; P ⳱ 0.0047, by chi-square test for linear trend) was a very likely mode of transmission. Simi- larly, higher levels of child-parent communication were asso- ciated with higher levels of correct knowledge of Ae. aegypti larval habitats (regression; P ⳱ 0.008), but higher levels of incorrect knowledge of cause of dengue among parents of elementary school children (regression; P⳱ 0.001). This sug- gests that while children may be learning at school what does not cause dengue, they are not effectively communicating this

information to their parents. Exposure to posters was slightly higher among the parents of elementary school children exposed to the program, and exposure to PSAs was uniform across all groups of parents (Table 1).

Impact on behavior: comparison of larval infestation in- dices. The overall Breteau Index was 27.5 for Ae. aegypti (for 703 inspected premises in the five municipalities treated and untreated) and 45.0 for all mosquitoes. For Ae. aegypti, the most numerous larval container habitats were intentional water storage containers (30.6% of all positive containers), plant-related containers (22.3%), tires (10.4%), small disposable containers (8.3%), and animal drinking dishes (7.3%).

Although dengue and Ae. aegypti are oftenconsidered urban pests, respective larval indices were higher as housing became more rural. The Ae. aegypti Breteau Index reached a high of 69.2 for parcelas (medium-sized rural lots) and 110.0 for fincas (rural estates). The elementary school program was associated with a significantly lower Breteau Index in one (Moca) of the two treated municipalities (Table 2). No effect on either index was detected for the Head Start program. The latter could be partly due to the small numbers of larval surveys conducted in the program’s treated group (see Discus- sion).

Impact on specific behavior: numbers of disposable con- tainers on house lots. The mean number of disposable con- tainers found per premise in the four study groups ranged from sevento 26, with very high variability from house to house. Adjusted (geometric) means for treated and untreated areas were not significantly different in the Head Start (5.2 versus 5.0, respectively) and elementary school programs (8.5 versus 7.9, respectively). The analysis therefore found no evi- dence that program activities decreased the number of disposable containers on people’s premises. A partial explana- tion for this was that both treated and untreated groups were already using dependable refuse collection services (see Dis- cussion); the numbers of disposable containers found were small and any differences had to be large to be detected by the study.

Impact on specific behavior: tires protected from mosqui- toes. The meannumber of tires per premise varied by study site, ranging from 0.28 tires per premise in the Head Start exposed group to 0.92 tires per premise inthe elementary school exposed group. Only a small portion of tires were adequately protected in the four study groups, ranging from 12% to 31%, but there was a statistically significant increase in the proportion of tires protected with increasing parent- child communication about dengue (Table 2).

Impact on specific behavior: protection of water storage containers from mosquitoes. The meannumber of water stor- age containers ranged from 1.92 to 1.37 per premise in the four study groups. There was a significant (two-thirds) reduction in mosquito-infested containers associated with exposure to the school intervention (Table 2), which was partially ex- plained by a strong protective effect of exposure to posters (Table 2).

Impact on specific behavior: use of aerosol insecticides.

The various components of the programs stress reduction of mosquitoes at their source: the larval habitats. Since it is im- possible at this time to eliminate all larval habitats, the programs also promote control of adult mosquitoes. Almost all educational materials mentioned spraying aerosol insecticides inside the house. Insecticide use, assessed through a proxy for

(6)

actual behavior, “Canyou show me a canof insecticide?,”

showed a positive associationwith the parent’s exposure to posters (Table 2).

DISCUSSION

The pilot community-based programs established by the Dengue Branch of CDC, in collaboration with the Puerto Rico Department of Health, showed significant positive impact on knowledge and behavior related to dengue prevention.

In terms of increased knowledge among children, the Mu- seum exhibit on Ae. aegypti was as successful as the school and Head Start-based programs. While it was not determined whether such short-term exposure for the child results inany knowledge and behavior increases for the parent, this is, to our knowledge, the first evaluationof a health exhibit ina children’s museum, and these favorable results should en- courage children’s museums and educational centers that stage hands-on health-related exhibits.^11–14

For parents, exposure to the elementary school program was associated with a significantly lower Ae. aegypti Breteau Index and a higher percentage of water storage containers protected from mosquitoes. The lack of associationobserved between the parents’ exposure to the Head Start dengue prevention program and residential infestation or key behaviors could be at least inpart due to the smaller sample size ac- quired from the treated household group, both for knowledge and exposure, as well as larval surveys. This reduced sample size was occasioned by unanticipated delays in the parental consent and data collection process during a narrow window of opportunity, as the end of the school year approached.

Increased communicationbetweenexposed childrenand their parents was associated with a higher proportion of tires protected from rain. Exposure to posters was associated with an increased proportion of water storage containers protected from mosquitoes and increased indoor use of aerosol insecticides. The effect of posters could be partially due to their increased use by program managers as a tool for interpersonal contact with parents. The exposure to PSAs, on the other hand, which were only aired by the mass media (seven TV stations on the island), was uniform across all groups and independent of treatment status (Table 1). The parent’s degree of exposure to PSAs was not associated with knowledge or behavior, but PSAs and radio spots are generally intended to complement inter-personal communication channels by raising general awareness rather than teaching specific skills needed for behavior change to occur.¹⁵

This analysis may have underestimated the pilot programs’

net impact on knowledge or behavior. There may be no true untreated group in Puerto Rico because dissemination of in- formationondengue preventionvia the mass media has occurred since 1963.³Island-wide campaigns and their intensity have coincided with large epidemics,^16,17and educational activities in one form or another have occurred almost continu- ously since 1985.^6,18

The effectiveness of larval source reduction is traditionally assessed through comparisonof Ae. aegypti larval indices, yet this presents many difficulties, especially when changes in vector infestation are used as a proxy for changes in human behavior. We therefore developed more sensitive analytical TABLE2 ImpactofdengueprogramsonAedesaegyptiresidentialinfestationandkeypreventivebehaviors* Unitof analysis

Comparisonfortheelementaryschoolprogram†Comparisonwithall‡parents’degreeofexposureto Treated groupUntreated groupStatistical procedureCommunication withchildPostersPSAsStatistical procedure Premise(n⳱242)(n⳱232)(n⳱332premises) Multiplelinear regressionMultiplelinear BreteauIndex§15.727.2P⳱0.045forMocaonlyNSNSNSregression Multiplelogistic regression HouseIndex¶12.4%17.7%NSNSNSNSMultiplelogistic regression TiresprotectedTire(n⳱223)(n⳱94)Chi-squaretestOR#⳱3.23(n⳱278premises)Chi-squarefor fromrain31%24%NSP⳱0.00047NSNSlineartrend ContainerswithWaterstorage(n⳱465)(n⳱438)(n⳱886waterstoragecontainers) anylarvaecontainer**2.6%8.0%P⳱0.00025NSOR⳱0.38 P⳱0.011NS AerosolinsecticideParent(n⳱140)(n⳱132)Multiplelogistic(n⳱342parentrespondents)Multiplelogistic observedstored onpremiserespondent51.4%43.2%regression NSNSOR⳱2.19 P⳱0.0059NSregression *PSAs⳱PublicServiceAnnouncements;NS⳱notsignificant;OR⳱oddsratio. †NotevaluatedfortheChildren’sMuseumexhibitonAe.aegyptiandnotshownfortheHeadStartprogram. ‡All⳱parentsofelementaryschoolandHeadStartchildren. §NumberofcontainersfoundtocontainAe.aegyptilarvaeforevery100premisesinspected. ¶PercentinspectedpremisesfoundtocontainAe.aegyptilarvae. #Oddsratioswerecalculatedfor6–8pointsofexposureoverthebaseline(0–1pointsofexposure). **Includes55-gallondrumsandbucketsandbottlesofvarioussizes.

(7)

methods, such as number of containers per premise positive for any type of larvae rather than for Ae. aegypti only, number of protected tires rather than simply the dry ones, and number of all disposable containers per premise rather than wet or positive only, and we included a time variable to account for changes in climatic conditions during the survey and urban/

rural housing variables (housing type and patio size) in the analysis. Lastly, analyses by degree of exposure rather than simply by treatment group gave a better indication of how the programs’ impact onbehavior was delivered.

With the detectionof dengue-3 virus in1998, all four dengue serotypes are now circulating in Puerto Rico, raising con- cerns about further epidemic transmission in the near fu- ture.¹⁹The programs presented here have had a greater impact onchildren’s knowledge about dengue thanonpromoted humanbehaviors and mosquito infestationlevels. Inpart, this is to be expected because the direct participants of these programs are school and pre-school aged children who cannot be expected to fully control their surrounding environment.

Most parents have sufficient knowledge about dengue (Table 1), but do not seem to follow through on the day-to-day ac- tions that they recognize as useful. Therefore, to improve the effectiveness of a behavior change strategy for dengue control inPuerto Rico, priority should be givento providing information and training on how to locate and control potential container habitats.

Efforts oftenfocus onclean-up campaigns of disposable containers. However, since refuse collection is fairly dependable in Puerto Rico, with 89% of the respondents to the parents’ survey stating that their refuse is always picked up, containers considered disposable by dengue experts must not be disposable for citizens. In this study, the disposables were a relative minority, compared with the number of water storage and plant related containers. In view of the recurrent prob- lems with the water supply system experienced in some areas of the island, there is a clear need to provide information and training on how to store water safely. Water storage containers have become the most important larval container habitats as a result of water rationing in some sectors of Puerto Rico during recent years. Many of the materials used in the dengue programs did not address that problem at the time of the evaluation (but now reflect this reality). Since many of the containers, especially the plant-related ones, are either hid- den, far from the house (in this study, 36.8% of the plant- related containers were “invisible” and more than 15 feet away from the house), support activities need to teach specific skills on how to locate and control those as well.

Mass media (such as PSAs) and community-wide events should provide appropriate cues to practice specific behaviors on a routine basis and not just during epidemics, while setting- specific programs (such as school- and Head Start-based) should provide positive feedback to those who are performing the target behaviors and should promote the benefits of performing the behaviors to those who are not, as in community- based programs to promote physical activity, for example.^20,21 Data from entomologic and epidemiologic surveillance needs to be fed back to the communities from where it is collected to let people know what the impact of control measures has been. Finally, educational messages should emphasize that the use of insecticides, by the individual or by government, is not an alternative for source reduction.

Numerous studies have confirmed the findings of the cur-

rent study that increasing awareness of the seriousness of dengue and DHF and prevention methods does not translate into behavior change, as indicated by decreases in larval indices.^22–26Some program planners have reasoned that involving schoolchildren may be a way to bridge this gap. Involve- ment of children is postulated to promote behavior change in parents, as well as to introduce the children to the concepts of dengue prevention at an early age. Unfortunately, there are no reports of controlled trials of the independent effect of the involvement of children on the effectiveness on dengue pre- ventioninthe literature, although there is a number of de- scriptions of programs where schoolchildren have partici- pated.^18,27–29The three main program components evaluated in this study all sought to change the behavior of parents through involving them in school-based dengue activities or through communication of dengue messages to them by their children. Takenas a whole, the results of this evaluationsug- gest that schoolchildren do communicate with their parents about dengue prevention, and that school programs can increase parental involvement in dengue control, but more specific messages about the behaviors to be performed need to be directed directly at parents.

Acknowledgments: We thank Hilda Seda for her multiple input during various stages of the study; the Puerto Rico Department of Health, especially the Division of Community Hygiene (Higieniza- cio´ndel Ambiente Fı´sico Inmediato); the Puerto Rico Department of Education, especially the elementary schools of Moca, Aguadilla, and Cidra; all Puerto Rico Head Start grantee organizations, especially servicios al niño y de sarrollo comunal (SENDEC), Municipality of Carolina and Servicios a la Familia; and the Children’s Museum in Old San Juan for their interest in participating in and facilitating our work; the students of the University of Puerto Rico who assisted with the data collection; as well as the numerous parents and children who welcomed us to their homes.

Financial support: This study was supported by CDC contract 200- 93-065.

Authors’ addresses: Peter J. Winch, E-mail: pwinch@jhsph.edu, and Elli Leontsini, E-mail: eleontsi@jhsph.edu, Department of Interna- tional Health, Bloomberg School of Public Health, The Johns Hop- kins University, 615 North Wolfe Street, Baltimore, MD 21205-2179, Telephone: 410-955-3859, Fax: 410-614-1419. Jose´ G. Rigau-Pe´rez and Gary G. Clark, Dengue Branch, Division of Vector-Borne Infec- tious Diseases, Centers for Disease Control and Prevention, 1324 Calle Can˜ada, San Juan, PR 00920-3860, Telephone: 787-706-2399, Fax: 787-706-2496. MervinRuiz-Pe´rez, WESTAT, 1650 Research Boulevard, Rockville, MD 20850-3195, E-mail: mruiz@westat.com.

Duane J. Gubler, Division of Vector-Borne Infectious Diseases, Cen- ters for Disease Control and Prevention, P.O. Box 2087, Fort Collins, CO 80522-2087, Telephone: 970-221-6462, Fax: 970-221-6476.

Reprint requests: Gary G. Clark, Dengue Branch, Division of Vector- Borne Infectious Diseases, Centers for Disease Control and Preven- tion, 1324 Calle Can˜ada, SanJuan, PR 00920-3860.

REFERENCES

1. World Health Organization, 1997. Dengue Haemorragic Fever:

Diagnosis, Treatment, Prevention and Control. Second edition.

Geneva: World Health Organization, 12–23.

2. Neff JM, Morris L, Gonza´lez-Alcover R, ColemanPH, Lyss SB, NegrónH, 1967. Dengue fever ina Puerto Ricancommunity.

Am J Epidemiol 86: 162–184.

3. Mun˜oz RA, Elinson J, Luck E, 1965. Knowledge of Dengue in Puerto Rico, February - April, 1964. Puerto Rico Master Sample Survey of Health and Welfare, Series 2, Number 1. San Juan, Puerto Rico: University of Puerto Rico, School of Medi- cine, Department of Public Health and Commonwealth of Puerto Rico, Department of Health.

4. Lo´pez-Correa RH, Cline BL, Ramı´rez-Ronda C, Bermúdez R,

(8)

Sather GE, Kuno G, 1978. Dengue fever with hemorrhagic manifestation: a report of three cases from Puerto Rico. Am J Trop Med Hyg 27: 1216–1224.

5. Gubler DJ, Clark GG, 1995. Dengue/dengue hemorrhagic fever:

the emergence of a global health problem. Emerg Infect Dis 1:

55–57.

6. Gubler DJ, Casta-Vélez A, 1991. A program for prevention and control of epidemic dengue and dengue hemorrhagic fever in Puerto Rico and the U.S. Virgin Islands. Bull Pan Am Health Organ 25: 237–247.

7. Rigau-Pérez JG, Vorndam AV, Clark GG, 2001. The dengue and dengue hemorrhagic fever epidemic in Puerto Rico, 1994–

1995. Am J Trop Med Hyg 64: 67–74.

8. Meltzer MI, Rigau-Pérez JG, Clark GG, Reiter P, Gubler DJ, 1998. Using disability-adjusted life years to assess the eco- nomic impact of dengue in Puerto Rico: 1984–1994. Am J Trop Med Hyg 59: 265–271.

9. Oficina de Presupuesto y Gerencia, 1992. Puerto Rico: Datos de Población por Municipio, Censo 1990. San Juan, PR: Oficina de Presupuesto y Gerencia de Puerto Rico.

10. Cronbach LJ, 1951. Coefficient alpha and the internal structure of tests. Psychometrika 16: 297–334.

11. Most BW, 1977. Children’s museum helps dentists. Cal 42: 2–3.

12. NewmanIM, 1979. Teaching health ina Natural History Mu- seum. J Sch Health 49: 104–108.

13. Beery QC, Loney J, 1986. Discovering the voice in the children’s museum. Int J Pediatr Otorhinolaryngol 11: 275–280.

14. Gunby P, 1996. ‘Hands-on’ center may offer a clue to health educationinthe 21st century. JAMA 276: 1289–1290.

15. Rogers EM, 1995. Diffusion of Innovations. Fourth edition. New York: The Free Press.

16. Morens DM, Rigau-Pérez JG, López-Correa RH, Moore CG, Ruiz-TibénEE, Sather GE, Chiriboga J, EliasonDA, Casta- Vélez A, Woodall JP, 1986. Dengue in Puerto Rico, 1977:

public health response to characterize and control an epidemic of multiple serotypes. Am J Trop Med Hyg 35: 197–211.

17. Dietz V, Gubler DJ, Ortiz S, Kuno G, Casta-Vélez A, Sather GE, Gómez I, Vergne E, 1996. The 1986 dengue and dengue hemorrhagic fever epidemic inPuerto Rico: epidemiologic and clinical observations. P R Health Sci J 15: 201–210.

18. Gubler DJ, Clark GG, 1994. Community-based integrated control of Aedes aegypti: a brief overview of current programs. Am J Trop Med Hyg 50(Suppl): 50–60.

19. Centers for Disease Control and Prevention, 1998. Dengue out-

break associated with multiple serotypes-Puerto Rico, 1998.

Morb Mortal Wkly Rep MMWR 47: 952–956.

20. King A, 1998. How to promote physical activity in a community:

research experiences from the US highlighting different com- munity approaches. Patient Educ Couns 33 (Suppl 1): S3–S12.

21. Blake SM, Jeffery RW, Finnegan JR, 1987. Process evaluation of a community-based physical activity campaign: the Minnesota Heart Health Program experience. Health Educ Res 2: 115–

121.

22. Gupta P, Kumar P, Aggarwal OP, 1998. Knowledge, attitude and practices related to dengue in rural and slum areas of Delhi after the dengue epidemic of 1996. J Commun Dis 30: 107–112.

23. Chiaravalloti Neto F, de Moraes MS, Fernandes MA, 1998. Re- sults of activities encouraging community participation in dengue control in an outlying neighborhood of Sao Jose do Rio Preto, Sao Paulo, and the relationship between the popu- lation’s knowledge and habits. Cad Saude Publica 14: 101–109.

24. Degallier N, Vilarinhos PT, de Carvalho MS, Knox MB, Caetano J, 2000. People’s knowledge and practice about dengue, its vectors, and control means in Brasilia (DF), Brazil: its rel- evance with entomological factors. J Am Mosq Control Assoc 16: 114–123.

25. Rosenbaum J, Nathan MB, Ragoonanansingh R, Rawlins S, Gayle C, Chadee DD, Lloyd LS, 1995. Community participation in dengue prevention and control: a survey of knowledge, attitudes, and practice in Trinidad and Tobago. Am J Trop Med Hyg 53: 111–117.

26. Swaddiwudhipong W, Lerdlukanavonge P, Khumklam P, Koon- chote S, Nguntra P, Chaovakiratipong C, 1992. A survey of knowledge, attitude and practice of the prevention of dengue hemorrhagic fever inanurbancommunity of Thailand. South- east Asian J Trop Med Public Health 23: 207–211.

27. Passos AD, Rodrigues EM, Dal-Fabbro AL, 1998. Dengue con- trol inRibeirao Preto, Sao Paulo, Brazil. Cad Saude Publica 14: 123–128.

28. Madeira NG, Macharelli CA, Pedras JF, Delfino MC, 2002. Edu- cationinprimary school as a strategy to control dengue. Rev Soc Bras Med Trop 35: 221–226.

29. Kay BH, Nam VS, TienTV, YenNT, Phong TV, Diep VT, Ninh TU, Bektas A, Aaskov JG, 2002. Control of Aedes vectors of dengue in three provinces of Vietnam by use of Mesocyclops (Copepoda) and community-based methods validated by en- tomologic, clinical, and serologic surveillance. Am J Trop Med Hyg 66: 40–48.