Cost effectiveness of Aedes aegypti control programmes: participatory versus vertical

Transactions of the Royal Society of Tropical Medicine and Hygiene (2007) 101, 578—586

available at www.sciencedirect.com

journal homepage: www.elsevierhealth.com/journals/trst

Cost effectiveness of Aedes aegypti control programmes: participatory versus vertical A. Baly a,∗, M.E. Toledo a, M. Boelaert b, A. Reyes a, V. Vanlerberghe b, E. Ceballos c, M. Carvajal c, R. Maso c, M. La Rosa c, O. Denis c, P. Van der Stuyft b a

Instituto de Medicina Tropical ‘Pedro Kour´ı’, Department of Epidemiology, Autopista Novia del Mediodia Km 61⁄2, La Lisa, Ciudad de La Habana, Cuba b Epidemiology and Disease Control Unit, Institute of Tropical Medicine, Nationalestraat 155, 2000 Antwerp, Belgium c Provincial Center of Surveillance and Vector Control, Avenida Garzon, Santiago de Cuba, Cuba Received 13 June 2006; received in revised form 18 January 2007; accepted 18 January 2007 Available online 21 March 2007

KEYWORDS Dengue; Aedes aegypti; Control; Community-based programme; Cost-effectiveness analysis; Cuba

Summary We conducted an economic appraisal of two strategies for Aedes aegypti control: a vertical versus a community-based approach. Costs were calculated for the period 2000—2002 in three pilot areas of Santiago de Cuba where a community intervention was implemented and compared with three control areas with routine vertical programme activities. Reduction in A. aegypti foci was chosen as the measure of effectiveness. The pre-intervention number of foci (614 vs. 632) and economical costs for vector control (US$243 746 vs. US$263 486) were comparable in the intervention and control areas. During the intervention period (2001—2002), a 13% decrease in recurrent costs for the health system was observed. Within the control areas, these recurrent relative costs remained stable. The number of A. aegypti foci in the pilot areas and the control areas fell by 459 and 467, respectively. The community-based approach was more cost effective from a health system perspective (US$964 vs. US$1406 per focus) as well as from society perspective (US$1508 vs. US$1767 per focus). © 2007 Royal Society of Tropical Medicine and Hygiene. Published by Elsevier Ltd. All rights reserved.

1. Introduction Ineffective vector control, urbanisation and increased air travel have led to a resurgence of the mosquito Aedes

∗

Corresponding author. Tel.: +53 7 202 0652. E-mail addresses: [email protected], [email protected] (A. Baly).

aegypti in Latin America and to the emergence of dengue as a public health problem (Gubler and Casta-Valez, 1991). At present, vector control and early detection and control of outbreaks are the only strategies available to reduce the impact of dengue (Gubler and Casta-Valez, 1991). Ultimately, prevention of dengue epidemics will depend on longterm vector control. The successful A. aegypti control campaigns in the Americas during the 1950s and 1960s had, by 1972, achieved eradication of this mosquito in 21 countries

0035-9203/$ — see front matter © 2007 Royal Society of Tropical Medicine and Hygiene. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.trstmh.2007.01.002

Cost effectiveness of Aedes aegypti control programmes within the region. However, by the end of the 1970s the majority of countries suffered re-infestation because financial support for control programmes had dwindled over time and, in some cases, was even abandoned (Cruz, 2002). It is commonly accepted that participation of the community in vertical control programme activities is critical to achieve sustainable and cost-effective vector control (Gubler, 1989; Liborio et al., 2004). No data exist on the total expenditure committed to dengue prevention and control in the Americas. This can be explained by the absence of a unified control structure and because manpower and supplies are usually not exclusively committed to dengue control. Typically, during epidemics, expenditure by governments and international donors increases dramatically, especially on insecticides, whereas between epidemics there is little money or resources available for routine control operations (PAHO, 1994). It has been estimated that in 25 endemic countries, only US$331 million and US$671 million was spent in 1996 and 1997, respectively, on routine vector control (PAHO, 1999), when the required budget was some US$1.6 billion. There is also little evidence on the cost effectiveness of larval control (McConnell and Gubler, 2003) and of recently promoted strategies for dengue health promotion that adopt a behaviorist approach to achieve behavioural changes (Parks and Lloyd, 2004). This study presents a cost-effectiveness analysis of two alternative strategies for A. aegypti control: a classical vertical vector control programme and a community-based intervention integrated in the vertical programme. Whereas the former is managed in a top-down manner, with planning and managerial decision-making all centralised at the national level, the latter involved local communities in the planning and management of different control activities at the local level (Toledo et al., 2007). Both strategies were implemented in parallel in Santiago de Cuba, Cuba, during 2001 and 2002.

2. Materials and methods 2.1. Background Dengue in Cuba occurs in outbreaks (Kouri et al., 1989; Pelaez et al., 2004; Valdes et al., 1999), which must be controlled at considerably greater cost than that of routine control programmes. Costs of control measures for the 1981 nationwide outbreak and the 1997 outbreak in Santiago de Cuba were estimated at, respectively, US$103 million and US$27 million (Guzman et al., 1992; Valdes et al., 2002). By 1981, the Cuban government had already launched a nationwide A. aegypti control programme managed by the Ministry of Health (MoH). Campaign workers at the primary healthcare level perform the main vector control activities. These procedures comprise: source reduction through periodic inspection of houses; adding larvicides to water storage containers; selective perifocal spraying (application of adulticide in a radius of 100—300 m around the detected foci) against adult mosquitoes; quality control; health education; and enforcement of mosquito control legislation (Toledo et al., 2007).

579 Despite high-level political support, adequate financial resources and high coverage, the programme did not manage to sustain reductions in vector infestation levels across all areas (house index <1%). To highlight a specific example, by 1987 the eastern province of Santiago de Cuba had succeeded in eradicating the vector but by 1992 began to suffer from re-infestation (Valdes et al., 1999). Difficulties with the reliability of the water supply led the population to store water in all kinds of containers, many of them in poor condition or without covers. Although campaign workers treated these vessels with larvicides, constant cleaning soon negated the effect and rendered the procedure pointless and wasteful (de la Cruz et al., 1999).

2.2. Study site A cost-effectiveness study was conducted in the city (or municipality) of Santiago de Cuba. Santiago de Cuba boasts a population of 470 000 inhabitants who are cared for by a dense network of family medical practices staffed by family doctors and nurses organised into nine health areas. In 2000, the provincial health authorities decided to develop and test a strategy by introducing community participation into the routine vector control programme. Twenty family medical practices within three health areas suffering similar A. aegypti infestation levels were randomly selected for this novel approach. An identical number of control practices were earmarked within three comparable areas of the same municipality. A full description of the intervention study is given elsewhere (Toledo et al., 2007).

2.3. Study type, time horizon and description of strategies A cost-effectiveness analysis was carried out from a number of different perspectives: the health system provider (establishing the total provider cost for vector control as well as the contribution from the primary healthcare services); the vertical programme (provider cost of vector control only); a community perspective (estimating the value of unpaid time given by all actors at the community level, including unpaid time contributed by health and vector control personnel); and the society (health system and community). Both the time and analytical horizon ran for 2 years (2001—2002), which coincided with the implementation period of the community-based intervention scheme. A full description of the two strategies follows. 2.3.1. Vertical vector control programme (control areas) The national programme carried out regular A. aegypti control activities throughout 2000. From early 2001, these activities were stepped up in light of a dengue outbreak in Havana. Intensified focal and perifocal larval control, blanket spraying for mosquito adulticiding and replacement of defective water tanks were among the measures taken. The intervals between house inspection cycles required for larval control were reduced from 22 days to 7—11 days. Finally, local leaders were trained in the delivery of MoHsponsored information, education and communication (IEC) advice regarding dengue and the need to promote environmental risk reduction (Toledo et al., 2007).

580 2.3.2. Community participation strategy (intervention areas) An alternative dengue control strategy was introduced at the end of 2000 to encourage active community participation in disease prevention in the regular A. aegypti control programme. An external research group from the Institute of Tropical Medicine ‘Pedro Kour´ı’ in Havana, in co-ordination with the health authorities of Santiago set up a local researcher team that was to become in charge of guiding the project. The external and local teams jointly designed the intervention, based on formative research, through a participative process (Toledo et al., 2006). The main goal of the intervention was to mobilise the community to become involved in all stages of A. aegypti control, from problem identification through planning and implementation and up to the final evaluation. A decision was taken to use the primary healthcare network as the vehicle for this new initiative. In early 2001, the family medical practices involved decided to create their own community working groups (CWG). Across the whole intervention area, 20 such CWGs were created, one for each community that is served by a family medical practice. The strategy was truly community based and can be assimilated to other ‘community-directed’ initiatives. Although the CWGs were organised around the family doctors’ practices, the ‘direction’ given by the health services was limited to the sharing of information about the importance of dengue control. Once the CWG was formed, including formal and informal community leaders and volunteers as well as interested medical and vector control staff, complete autonomy was granted to the group to establish priorities and to devise action plans. No financial incentives were offered to members. They identified main health problems and needs, and elaborated, implemented and evaluated their action plans, which varied widely from the transformation of garbage belts into vegetable gardens to repairing of broken water pipes, sealing of basements and manufacturing of lids for water containers. The necessary equipment and materials were provided free of charge by the local government, whilst members of the community gave their labour for nothing. All these activities were accompanied by a locally designed public health education strategy, which aimed to mobilise the population and to promote healthy behaviour (Toledo et al., 2007).

2.4. Data collection and analysis 2.4.1. Costs The economic costs of both strategies were estimated for the fiscal years 2000 (before the intervention), 2001 and 2002 (during the implementation). The sources of cost and activity data were the Cuban accountancy system for health ´nica del sistema de salud) and the (CONUS-Contabilidad u Cuban health information system. Cost inputs were itemised and allocated according to the WHO methodology (Johns et al., 2003). Because of the labour intensive nature of both the classical control programme and the complementary community-based activities, the principal costs are virtually all recurring, including those that relate to the salaries, consumables and notional hourly pay for time actually given free by the community.

A. Baly et al. The labour costs of the campaign workers both in control and intervention areas were derived from the bookkeeping records maintained in each of the health areas. They included the wages paid to the workers involved in larval control, adult mosquito control and quality control and to the supervisors of those activities. The share of the salary paid to primary healthcare workers (doctors, nurses, epidemiologists, technicians, health managers), who gave part of their time to the vector control programme or to the community participation strategy, was also included. Semistructured interviews were carried out to estimate the proportion of staff time allocated to each type of activity. The costs of consumables used for larval control and spraying were determined by multiplying the average value of products used per house for each of these activities by the number of houses inspected or sprayed in a single year. These data were collected by non-participant direct observation of vector control activities by trained personnel on the one hand and from accounting departments and the epidemiological information systems on the other hand. The costs of training and social communication were estimated from the respective reports and registers of the vertical programme and the intervention project. They include cost of materials, salaries of teachers and other personnel and estimated cost for trainee time based on the government study allowance rate. Operating expenses included food, travel allowances, fuel and lubricants for vehicles, electricity and repair of spraying equipment. These costs were estimated from direct observation, interviews and from an examination of the bookkeeping records of the provincial and municipal accountancy departments. Capital costs included vehicles and equipment. Annual capital goods depreciation was calculated using the linear method: 5 years of useful life and a discount rate of 6%. The cost of buildings was not included because the information was not available and a local market does not exist in Cuba to consider replacement costs. In any event, the depreciation cost of these capital items relative to the total cost of the programme is very small because of the programme’s labour intensive nature. The value of unpaid community work was estimated by valuing it at the same rate that a similar type of employment would be reimbursed in the government sector. This rate was also applied to the additional time that health professionals volunteered to work over their contractual hours in, for example, the elimination of environmental risks (potential breeding site of any kind) and co-operation with campaign workers, intersectoral co-ordination and programme monitoring. To estimate the value of the ‘community contribution’, healthcare personnel, campaign workers and health promoters together with members of households were surveyed and interviewed. In addition, community action plans and reports drawn up by health promoters and local leaders were reviewed. The cost of free-of-charge equipment and materials provided by local government was not included. Equipment is not rented by the government, therefore capital depreciation because of the time used was almost zero. The kind of material supplied varied substantially from community to community, but overall the amounts were very small, so it was decided not to include them in the calculations.

Cost effectiveness of Aedes aegypti control programmes The cost per item was calculated both for the baseline (2000) and for the intervention period (2001—2002, annual and cumulative) and was also expressed as a percentage of the total costs. The costs in Cuban pesos were standardised at 2002 prices using a GNP implicit deflator (Comisi´ on Econ´ omica para Am´ erica Latina y el Caribe, 2004) and were then converted to $US at the official exchange rate (1 peso = $US1). The total economic cost to the society, the cost to the health system and to the vector control programme (including larval control, which includes inspection and larvicide application, larval quality control, larval operational cost, training and administrative share) and cost per inhabitant (p.i.) covered were computed. Inputs were determined where the intervention had a higher or lower cost compared with the control area. The absolute and percentage cost differences between the two strategies, by each cost input estimated, were calculated using the procedure proposed by Reynolds and Gaspari (1986). 2.4.2. Effectiveness The intervention study (Toledo et al., 2007) used several different measures to analyse effectiveness: entomological indicators extracted from routine data collection (number of foci, house index, changes in location of main breeding sites) and behavioural change indicators (numbers of correctly covered tanks, unprotected artificial containers and water containers protected by larvicide). The reduction in number of foci, defined as any kind of container containing a larval stage of A. aegypti, was used as the effectiveness measure. The reduction was calculated for three discrete periods: 2001 vs. 2000 (first year of intervention), 2002 vs. 2001 (second year of intervention) and 2002 vs. 2000 (whole period of intervention). 2.4.3. Cost effectiveness Cost effectiveness was calculated for three periods (2000—2001, 2001—2002 and 2000—2002). The overall annual costs were divided by the reduction in the number of foci for the corresponding period. Incremental cost

581 effectiveness for 2000—2002 was calculated by dividing the difference in total cost by the difference in overall reduction in the number of foci between the two strategies.

3. Results 3.1. Cost analysis The greater part of the budget (60%) committed by the municipality as a whole to the vertical programme was spent on wages (Table 1) owing to the labour intensive character of the work. Following management decentralisation in 2001, a portion of the capital stock was transferred to other governmental agencies, hence the substantial decrease in the value of fixed assets from the previous year. From 2001, high vector infestation levels led to shorter programme cycles (the time interval between house inspections); from 22 days to initially 11 days and subsequently 7 days, with a concomitant increase in the work force employed. There was a corresponding increase in total salary cost; consequently the p.i. cost grew from $US13 in 2000 to $US24 in 2002. The intervention and control areas were comparable with regard to the number of houses, infestation level, number of inhabitants per house and environmental risk (Table 2). The figures in Table 3 show that the activities of the vertical control programme were similar in both areas, leading to comparable expenditure during the baseline period. Furthermore, if the costs of the primary healthcare team and the community contribution are excluded, the relationship (expressed as a percentage) between recurrent and capital costs and the total cost are similar to those reported in Table 1. This corroborates our baseline cost estimations. The accounts show that there was an absolute increase in recurrent costs in both areas between 2001 and 2002. However, within the intervention area the share of recurrent costs relative to the total cost decreased from 76.3% to 63.7%. There was also a shift from financial to economic costs because community costs increased from 23.5% to 36.1%. Within the control area, the cost breakdown by item remained the same throughout the study period (Figure 1).

Table 1 Total cost (US$a ) of the vertical Aedes aegypti control programme for the whole municipality of Santiago de Cuba, 2000—2002b Item

2000

2001

2002

Total

Recurrent costs Personnel Training Supplies Communications Transport Operative costs

6 212 607 4 685 441 25 661 894 762 15 468 68 516 522 758

7 066 096 4 268 208 20 530 668 013 19 868 882 470 1 207 008

11 031 153 5 623 337 12 399 1 011 693 33 046 929 944 3 420 733

24 309 856 14 576 986 58 590 2 574 467 68 382 1880 929 5 150 499

99.7 59.8 0.2 10.6 0.3 7.7 21.1

57 265

13 368

14 555

85 188

0.3

6 269 872 13

7 079 461 15

11 045 706 24

24 395 039 52

Capital depreciation Total Cost per inhabitant

Source: Accountancy of the Health Department (municipality level). a US$ constant 2002. b All numbers were rounded up.

%

100 —–

582 Table 2

A. Baly et al. Characteristics of intervention and control areas before the intervention, Santiago de Cuba, 2000

Baseline information

Intervention area

Control area

Difference (95% CI)

No. of houses No. of house blocks Average no. of subjects per house Average no. of containers for water storage per house Yearly no. of Aedes aegypti foci detected Principal breeding site

2400 48 4.8 8.5

2600 49 4.2 7.9

— — 0.6 (−0.5 to 1.2) 0.59 (−0.1 to 1.3)

614

632

—

Ground tanks for water storage (indoor) 0.20 (0.1—0.37) 1.23 (0.7—2.6)

Ground tanks for water storage (indoor) 0.3 (0.28—0.34) 2.08 (1.91—2.43)

—

45.6%

55.2%

9.7% (−0.8 to 19.8)

61.9%

60.0%

1.9% (−8.1 to 12.0)

70.0%

69.6%

0.4% (−10.0 to 9.0)

Median container index (95% CI) Median house index (95% CI) Main behavioural risk factors Houses with incorrect use of larvicide Houses with unprotected artificial containers Houses with badly covered containers for water storage *

* *

No differences, as overlap of confidence intervals.

Whilst p.i. costs were comparable prior to the intervention (2000), by 2002 the economic cost p.i. had reached $US32 in the intervention area and $US41 in the control area. In the same year, the financial costs to the health system, vertical programme and larval control programme were $US20 p.i., $US16 p.i. and US$7 p.i., respectively. In the control zone these costs were higher: $US32, $US28 and $US12, respectively (Figure 2). The additional cost to the health system was US$214 198 (48% higher) in the control area compared with the intervention area (Table 4). All provider cost items, except for social communication costs, were higher in the control area. However, the cost to the community was 48.1% higher in the intervention area.

Table 3

Table 5 illustrates the amount of time invested in the intervention area by the different kind of actors involved during 2 years of intervention. Most of the time was invested in surveillance of intradomiciliary and extradomiciliary risks (132 600 h; 36.3%), followed by environmental risks (100 830 h; 27.6%), community sanitation (65 080 h; 17.8%) and finally planning and evaluation of community activities (39 912 h; 10.9%).

3.2. Effectiveness and cost effectiveness At baseline, intervention and control areas reported 614 and 632 foci, respectively. These numbers decreased in 2001

Economic costs (US$a ) per year in the intervention and control areas, Santiago de Cuba, 2000—2002b

Input

Intervention area Baseline (2000)

Running cost/year 2001

2002

210 129 92 36 16 15

230 154 113 41 16 12

441 283 206 77 32 27

053 753 067 686 499 812

202 127 101 26 19 5

943 706 188 518 542 877

Running cost/year 2001

2002

296 189 141 48 24 6

358 236 180 55 30 11

512 932 664 268 579 845

064 117 349 768 794 129

Total 2001—2002

45 181 458 57 303

50 189 477 113 318

46 801 952 136 490

96 990 1 429 249 808

49 818 468 60 075

75 157 749 94 552

80 025 1 354 74 077

155 182 2 103 168 629

Total costs

243 746

324 509

367 782

692 290

263 486

391 813

433 496

825 309

US$ constant 2002. All numbers were rounded up.

340 699 182 518 422 417

Baseline 2000

185 117 90 26 17 6

b

714 053 885 168 077 394

Total 2001—2002

Recurrent costs Personnel Vector campaign Primary health care Supplies Training and social communication Operating costs Capital costs Community costs

a

985 289 471 818 354 161

Control area

654 426 322 104 55 17

577 049 013 036 373 973

Cost effectiveness of Aedes aegypti control programmes

Figure 1 Relative costs (US$ constant 2002) at baseline and during the implementation period in the intervention and control areas, Santiago de Cuba, 2000—2002.  Recurrent cost;  capital cost;

community cost.

583

Figure 2 Cost per inhabitant (US$) and per year in the intervention and control areas, Santiago de Cuba, 2000—2002.  Economic cost;  health system cost; cost;

to 272 and 274, respectively, and further in 2002 to 155 and 165. Overall during the intervention period (2001—2002) both zones showed a similar reduction in number of foci: by 459 in the intervention area and 467 in the control area (Table 6). Community-based intervention was more cost effective from the point of view of society as a whole (Table 6). It was also more cost effective both from the health system and the vertical programme perspective. In both areas, the cost per focus eliminated was lower in 2001 compared with 2002. From the point of view of the health system, vertical programme and society (health system and community), the incremental cost of eliminating one additional focus was much higher in the control area than in the intervention area. However, the cost was concomitantly lower from the community point of view in the control area.

vertical programme

larval control cost.

4. Discussion This study shows that in Santiago de Cuba the dengue control programme integrating a community-based intervention strategy was more cost effective than an intensive vertical programme alone. From an overall health system perspective, the former appears to be a good investment. However, policy-makers should realise that community involvement in vector control is not a ‘free ride’ and carries a substantial opportunity cost in terms of volunteer time spent on control. It is vitally important to acknowledge the substantial investment in health made by the community in the form of unpaid labour contributed to

Table 4 Absolute and percentage cost (US$a ) differences by each cost input estimated during the intervention period, Santiago de Cuba, 2001—2002

Inputs where the intervention costs less Recurrent costs Personnel Vector campaign Primary health care Supplies Larval control Spraying Operating costs Capital costs Vehicles Equipment Inputs where the intervention costs more Training and social communication Community costs Total a

US$ constant 2002.

Absolute difference between intervention and control areas

Percentage difference

−213523.8 −142296.3 −115946.3 −26350.0 −22874.0 −7382.5 −15491.5 −58192.0 −674.0 −623.0 −51.0

−48.4 −50.1 −56.3 −33.9 −70.4 −49.5 −88.1 −60.0 −47.2 −85.7 −7.3

+9838.5 +81179.0

+54.7 +48.1

−133018.8

−19.2

584

A. Baly et al.

Table 5 Average time (in hours) over 2 years per type of social actor for each type of activity, and total time invested in the community participation components of the Aedes aegypti control programme intervention, Santiago de Cuba, 2001—2002 Activity

Social actor Primary healthcare teama

Planning and evaluation of activities (bimonthly CWG meeting, community meeting) Surveillance of environmental risk (outdoor) Surveillance of risk behaviours (indoor) Community sanitation Intersectoral co-ordination Administration Research and trainingd Total

% Campaign workera

Health promotera

Community health workerb

Household memberb

Totalc

96

96

600

89.3

20

39 912

10.9

192

900

800

118.3

50

100 830

27.6

260

1 100

1 200

176.7

48

132 600

36.3

100 108 89 96

100 120 80 96

90 72 — —

40 — — —

65 080 17 006 4 299 5 462

17.8 4.7 1.2 1.5

18 137

8 988

90 140

209 800

365 189

54.5 120 88 74 38 124

100

CWG: community working group. a Health system workers. b Community actors. c Sum of all time used by all involved actors. d Local researchers, excluding research team from Instituto de Medicina Tropical ‘Pedro Kour´ ı’.

vector control activities. At baseline, both intervention and control zones in our study were directly comparable and there were no differential external influences for the duration of the study. Therefore, it is fair to assume that equivalent reductions in the number of foci were

solely attributable to relative effectiveness. Any attempt to evaluate interventions of this type presents a number of difficulties, such as measuring effectiveness, valuing fixed—variable costs (stepwise) and diminishing returns to scale.

Table 6 Cost-effectiveness ratios and incremental cost-effectiveness ratios for participatory and vertical Aedes aegypti control, Santiago de Cuba, 2001—2002 Perspective

Intervention area Total cost

Control area

Incremental cost per focus eliminated

Reduction in no. of foci

Cost-effectiveness ratioa

Total cost

Reduction in no. of foci

Cost-effectiveness ratioa

Health system 2001 211 191 2002 231 292 2001—2002 442 483

342 117 459

617 1 977 964

297 261 359 419 656 680

358 109 467

830 3 297 1 406

26 775

Vertical programme 2001 175 023 2002 189 774 2001—2002 364 796

342 117 459

512 1 622 795

248 993 303 651 552 644

358 109 467

696 2 786 1 183

23 481

Community 2001 2002 2001—2002

113 318 136 490 249 808

342 117 459

331 1 166 544

94 552 74 077 168 629

358 109 467

264 680 361

−10 147b

Society 2001 2002 2001—2002

324 509 367 782 692 291

342 117 459

949 3 143 1 508

391 813 433 496 825 309

358 109 467

1 094 3 977 1 767

16 628

a b

US$ per eliminated focus (US$ constant 2002). In favour of the control area.

Cost effectiveness of Aedes aegypti control programmes In spite of the intense vector control programme over 20 years, the country has not been exempt of introduction of virus of dengue. Nevertheless, the active vigilance and the established control have permitted detecting prematurely the virus’s presence, avoiding big outbreaks and that dengue become an endemic disease (Guzman et al., 2006). Therefore, effectiveness of vector control interventions cannot be determined by follow-up of dengue-specific morbidity and mortality. Entomological indices (house, Breteau and container indices) are used as surrogate markers of epidemic risk, but the functional relationship between the index scores and the occurrence of dengue outbreaks is not well known. Moreover, their sensitivity when used to evaluate community-based strategies has been questioned (Kay and Vu, 2005). In our study, the most frequently found larval breeding sites were the groundwater storage tanks that are kept indoors or elsewhere on the premises close to the house. Their infestation is directly related to the behaviour of the population: whether they are kept uncovered and/or treated with larvicides (Toledo et al., 2007). A reduction in the entomological indices for these specific containers might possibly offer a better measure of the effectiveness of community-based interventions. However, only a small proportion of ground tanks registered positive at baseline in our study (in line with the general indices), making the unit cost of any reduction achieved unstable and difficult to interpret. Final outcome measures such as disability-adjusted life years or quality-adjusted life years are recommended for the cost-effectiveness analysis of health promotion programmes (Haddix and Teutsch, 2003). Ranking competing alternatives on the basis of cost-effectiveness ratios constructed from surrogate markers has been questioned, but in studies like ours where there is no alternative, their use is justified (Haddix and Teutsch, 1996, 2003). We have estimated that the 2002 financial cost of the A. aegypti control programme in Santiago amounts to approximately $US16 000 and $US28 000 per 1000 inhabitants for the intervention and control areas, respectively. Shepard et al. (2004) have recorded the costs of control programmes in several different countries analysed per 1000 inhabitants: $US15 in Indonesia (1998), $US81 and $US188 in Thailand (1994 and 1998), $US204 in Malaysia (2002) and $US2400 in Singapore (2000). For 17 Caribbean islands (1990) it ranged from $US140 to $US8490. Our estimates seem very high, but it is difficult to compare them with those previously reported since no information was given on the coverage or intensity of these programmes or the background epidemiological situation. McConnell and Gubler (2003) have demonstrated, based on mathematical modelling of dengue transmission data from Puerto Rico, that emergency larval control activities (without an early warning system) are more cost effective than doing nothing if ≤$US6 is spent p.i. In our intervention area, during 2002 the cost of the routine larval control was approximately $US7 p.i. (in the control area it reached $US12 p.i.). These numbers come close to the threshold proposed by McConnell, which suggests that the control activities implemented in Santiago were worthwhile. In our study, the incremental cost incurred by the vertical programme to eliminate one additional focus is very high compared with a community-based approach. This indicates

585 that from a health system perspective, the communitybased intervention can produce savings that might then be used to finance other control programme activities (dengue-related or not) or to address direct causes of vector proliferation such as water supply problems. Community-based strategies are generally difficult to implement and it often takes time before their impact becomes apparent. This discourages governments from investing time, money and human resources to develop such strategies (Winch et al., 1992). On the other hand, they are sometimes seen as attractive low-cost alternatives to vertical programmes (Ugalde, 1985). From the community’s point of view, we should consider the opportunity cost of using unpaid volunteers to implement certain programme activities. Moreover, both the community-based strategy and the vertical programme were less cost effective in 2002 compared with 2001, probably owing to diminishing returns of scale. Over time, both require greater and greater efforts to eliminate each additional focus, and the problem-solving capacity required either stretches community resources or the programme budget. This phenomenon may have a negative effect on sustainability. We conclude that the described community-based intervention for A. aegypti control, when intertwined with the vertical control programme, appears to be the superior strategy. Although entomological indices reported by the national control programme are very low in Cuba, dengue outbreaks have occurred with this low level of infestation (Pelaez et al., 2004; Sanchez et al., 2006). Therefore, these findings can be useful for health decision-making in terms of resource allocation for vector control programmes of the other countries. Whether this remains the case in the long run, in particular once Aedes infestation has been reduced to very low levels, must be addressed by future studies. Conflicts of interest statement The authors have no conflicts of interest concerning the work reported in this paper.

Authors contributions AB conceived the evaluation, participated in the fieldwork, analysed the data and drafted the manuscript; MET conceived and designed the intervention trial and participated in the fieldwork; EC, MC, RM, MLR and OD participated in the fieldwork and data collection; AR, MB, VV and PVdS participated in the design of the intervention, the data analysis and drafted sections of the manuscript; all authors participated in interpretation of the data, critically revised subsequent drafts of the manuscript and approved the final version. Three teams of researchers assisted in this study: the field team in Santiago de Cuba and technical support teams of the Epidemiology Departments of the Tropical Institute of Havana and of Antwerp. AB is guarantor of the paper.

Acknowledgements We gratefully acknowledge the role played by the health sector staff involved in the dengue prevention and control activities. We also thank the people of Santiago de Cuba who

586 participated in the study. The study was partially funded through the framework agreement between the Institute of Tropical Medicine and the Belgium Directorate-General for Development Co-operation, project 95900.

References Comisi´ on Econ´ omica para Am´ erica Latina y el Caribe, 2004. Cuba: Evoluci´ on econ´ omica durante 2003 y perspectivas para 2004. http://www.eclac.cl/ [accessed 14 November 2006]. Cruz, R.R., 2002. Strategies for controlling dengue and Aedes aegypti in the Americas. Rev. Cubana Med. Trop. 54, 189—201 [in Spanish]. de la Cruz, A.M., Figueroa, D., Chacon, L., Gomez, M., Diaz, M., Finlay, C.M., 1999. Knowledge, opinions, and practices regarding Aedes aegypti. Rev. Cubana Med. Trop. 51, 135—137 [in Spanish]. Gubler, D.J., 1989. Aedes aegypti and Aedes aegypti-borne disease control in the 1990s: top down or bottom up. Charles Franklin Craig Lecture. Am. J. Trop. Med. Hyg. 40, 571—578. Gubler, D.J., Casta-Valez, 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. Guzman, M.G., Triana, C., Bravo, J., Kouri, G., 1992. The estimation of the economic damages caused as a consequence of the epidemic of hemorrhagic dengue in Cuba in 1981. Rev. Cubana Med. Trop. 44, 13—17 [in Spanish]. Guzman, M.G., Pelaez, O., Kouri, G., Quintana, I., Vazquez, S., Penton, M., Avila, L.C., 2006. Final characterization of and lessons learned from the dengue 3 epidemic in Cuba, 2001—2002. Rev. Panam. Salud Publica 19, 282—289 [in Spanish]. Haddix, A., Teutsch, S. (Eds), 1996. A Practical Guide to Prevention Effectiveness: decision and economic analysis. CDC, Atlanta, GA, p. 100. Haddix, A., Teutsch, S. (Eds), 2003. A Practical Guide to Prevention Effectiveness: decision and economic analysis, second ed. Oxford University Press, New York, NY, pp. 162—163. Johns, B., Baltussen, R., Hutubessy, R., 2003. Programme costs in the economic evaluation of health interventions. Cost Eff. Resour. Alloc. 1, 1. Kay, B., Vu, S.N., 2005. New strategy against Aedes aegypti in Vietnam. Lancet 365, 613—617. Kouri, G.P., Guzman, M.G., Bravo, J.R., Triana, C., 1989. Dengue haemorrhagic fever/dengue shock syndrome: lessons from the Cuban epidemic, 1981. Bull. World Health Organ. 67, 375—380. Liborio, M., Tomisani, A.D., Moyano, C.B., 2004. Dengue prevention strategies: Rosario, Argentina. Rev. Bras. Epidemiol. 7, 311—327. McConnell, K.J., Gubler, D.J., 2003. Guidelines on the costeffectiveness of larval control programs to reduce dengue

A. Baly et al. transmission in Puerto Rico. Rev. Panam. Salud Publica 14, 9—16. PAHO, 1994. Dengue and dengue haemorrhagic fever in the Americas: guidelines for prevention and control. Pan American Health Organization, Washington, DC, Rep Sci. Pub 548. PAHO, 1999. A Blueprint for Action for the Next Generation: dengue prevention and control. Pan American Health Organization, Washington, DC. http://www.paho.org/english/hcp/hct/ hct-136-99.doc [accessed September 2005]. Parks, W., Lloyd, L., 2004. Planning Social Mobilization and Communication for Dengue Fever Prevention and Control: a step-by-step guide. World Health Organization, Geneva, WHO/CDS/WMC/2004.2. Pelaez, O., Guzman, M.G., Kouri, G., Perez, R., San Martin, J.L., Vazquez, S., Rosario, D., Mora, R., Quintana, I., Bisset, J., Cancio, R., Masa, A.M., Castro, O., Gonzalez, D., Avila, L.C., Rodriguez, R., Alvarez, M., Pelegrino, J.L., Bernardo, L., Prado, I., 2004. Dengue 3 epidemic, Havana, 2001. Emerg. Infect. Dis. 10, 719—722. Reynolds, J., Gaspari, K.C., 1986. Operation Research Methods: cost-effectiveness analysis. PRICOR, Monograph Series, Method Brochure, No. 2. Sanchez, L., Vanlerberghe, V., Alfonso, L., Marquetti, M.C., Guzman, M.G., Bisset, J., Van der Stuyft, P., 2006. Aedes aegypti larval indices and risk for dengue epidemics. Emerg. Infect. Dis. 12, 800—806. Shepard, D.S., Suaya, J.A., Halstead, S.B., Nathan, M.B., Gubler, D.J., Mahoney, R.T., Wang, D.N., Meltzer, M.I., 2004. Costeffectiveness of a pediatric dengue vaccine. Vaccine 22, 1275—1280. Toledo, M.E., Baly, A., Ceballos, E., Boelaert, M., Van der Stuyft, P., 2006. Participaci´ on comunitaria en la prevenci´ on del dengue: un abordaje desde la perspectiva de los diferentes actores sociales. Revista Salud P´ ublica de M´ exico 48. Toledo, M.E., Vanlerberghe, V., Baly, A., Ceballos, E., Valdes, L., Searret, M., Boelaert, M., Van der Stuyft, P., 2007. Towards active community participation in dengue vector control: results from action research in Santiago de Cuba. Trans. R. Soc. Trop. Med. Hyg. 101, 56—63. Ugalde, A., 1985. Ideological dimensions of community participation in Latin American health programs. Soc. Sci. Med. 21, 41—53. Valdes, L., Guzman, M.G., Kouri, G., Delgado, J., Carbonell, I., Cabrera, M.V., Rosario, D., Vazquez, S., 1999. Epidemiology of dengue and hemorrhagic dengue in Santiago, Cuba 1997. Rev. Panam. Salud Publica 6, 16—25 [in Spanish]. Valdes, L.G., Mizhrahi, J.V., Guzman, M.G., 2002. Economic impact of dengue 2 epidemic in Santiago de Cuba, 1997. Rev. Cubana Med. Trop. 54, 220—227 [in Spanish]. Winch, P., Kendall, C., Gubler, D.J., 1992. Effectiveness of community participation in vector-borne disease control. Health Policy Plan. 7, 342—351.