Improved equity in measles vaccination from integrating insecticide-treated bednets in a vaccination campaign, Madagascar
Amélioration de l’équité en matière de vaccination contre la rougeole par l’intégration de moustiquaires imprégnées d’insecticide dans une campagne de vaccination, à Madagascar
Mejora de la equidad en la vacunación contra el sarampión integrando mosquiteras impregnadas con insecticida en la campaña de vacunación, Madagascar
Abstract
enObjective To evaluate the effect of integrating ITN distribution on measles vaccination campaign coverage in Madagascar.
Methods Nationwide cross-sectional survey to estimate measles vaccination coverage, nationally, and in districts with and without ITN integration. To evaluate the effect of ITN integration, propensity score matching was used to create comparable samples in ITN and non-ITN districts. Relative risks (RR) and 95% confidence intervals (CI) were estimated via log-binomial models. Equity ratios, defined as the coverage ratio between the lowest and highest household wealth quintile (Q), were used to assess equity in measles vaccination coverage.
Results National measles vaccination coverage during the campaign was 66.9% (95% CI 63.0–70.7). Among the propensity score subset, vaccination campaign coverage was higher in ITN districts (70.8%) than non-ITN districts (59.1%) (RR = 1.3, 95% CI 1.1–1.6). Among children in the poorest wealth quintile, vaccination coverage was higher in ITN than in non-ITN districts (Q1; RR = 2.4, 95% CI 1.2–4.8) and equity for measles vaccination was greater in ITN districts (equity ratio = 1.0, 95% CI 0.8–1.3) than in non-ITN districts (equity ratio = 0.4, 95% CI 0.2–0.8).
Conclusion Integration of ITN distribution with a vaccination campaign might improve measles vaccination coverage among the poor, thus providing protection for the most vulnerable and difficult to reach children.
Abstract
frObjectif: Evaluer l’effet de l’intégration de la distribution des moustiquaires imprégnées d’insecticide (MII) sur la couverture vaccinale contre la rougeole à Madagascar.
Méthodes: Enquête nationale transversale pour estimer la couverture vaccinale contre la rougeole, à l’échelle nationale et dans les districts avec et sans l’intégration des MII. Pour évaluer l’effet de l’intégration des MII, un score de propension correspondant a été utilisé pour créer des échantillons comparables dans les districts avec et sans MII. Les risques relatifs (RR) et un intervalle de confiance (IC) à 95% ont été estimés par l’intermédiaire de modèles log-binomiale. Les rapports d’équité, définis comme le ratio de la couverture entre les ménages appartenant au quintile (Q) de richesse le plus bas et le plus haut, ont été utilisés pour évaluer l’équité dans la couverture vaccinale contre la rougeole.
Résultats: La couverture nationale de la vaccination contre la rougeole durant la campagne était de 66,9% (IC 95%: 63,0 à 70,7). Parmi le sous-groupe de score de propension, la couverture de la campagne de vaccination a été plus étendue dans les districts avec MII (70,8%) que dans les districts sans MII (59,1%) (RR = 1,3; IC 95%: 1,1 à 1,6). Parmi les enfants dans le quintile de richesse le plus pauvre, la couverture vaccinale était plus élevée dans les districts avec MII que dans les districts sans MII (Q1; RR = 2,4; IC 95%: 1,2 à 4,8) et l’équité pour la vaccination contre la rougeole était supérieure dans les districts avec MII (ratio d’équité = 1,0; IC 95%: 0,8–1,3) que dans les districts sans MII (ratio d’équité = 0,4; IC 95%: 0,2 à 0,8).
Conclusion: L’intégration de la distribution des MII dans une campagne de vaccination pourrait améliorer la couverture vaccinale contre la rougeole chez les pauvres; offrant ainsi une protection pour les enfants les plus vulnérables et difficiles à atteindre.
Abstract
esObjetivo: Evaluar el efecto de integrar la distribución de mosquiteras impregnadas con insecticida (ITNs) dentro de la cobertura vacunal del sarampión en Madagascar.
Métodos: Estudio nacional croseccional para calcular la cobertura vacunal del sarampión, a nivel nacional y en distritos con y sin integración de ITNs. Para evaluar el efecto de la integración de ITNs, se utilizó la correspondencia del puntaje de propensión para crear muestras comparables en distritos con ITNs y sin ITNs. El riesgo relativo (RR) y los intervalos de confianza 95% (IC) se estimaron mediante modelos de regresión logística binomiales. Las tasas de equidad, definidas como la tasa de cobertura entre los quintiles de menor y mayor riqueza (Q), se utilizaron para evaluar la equidad en la cobertura de la vacunación de sarampión.
Resultado: La cobertura nacional de vacunación del sarampión durante la campaña fue del 66.9% (95% IC 63.0–70.7). Entre el grupo de puntaje de propensión, la cobertura de la campaña de vacunación era mayor en los distritos con ITNs (70.8%) que en distritos sin ellas (59.1%) (RR = 1.3, 95% IC 1.1–1.6). Entre los niños dentro del quintil más pobre, la cobertura vacunal era mayor en los distritos con ITN que en los distritos sin ITNs (Q1; RR = 2.4, 95% IC 1.2–4.8) y la equidad de la vacunación de sarampión era mayor en los distritos con ITNs (ratio de equidad = 1.0, 95% IC 0.8–1.3) que en distritos sin ITNs (ratio de equidad = 0.4, 95% IC 0.2–0.8).
Conclusión: La integración de la distribución de ITNs con una campaña vacunal puede mejorar la cobertura de la vacunación de sarampión entre los pobres; por lo tanto el dar protección a los niños más vulnerables y difíciles de alcanzar.
Introduction
Improving equitable coverage with both measles vaccination and ITNs is a critical step towards reducing morbidity and mortality among children in developing countries (Roll Back Malaria Partnership 2005, 2008; United Nations 2011). One strategy to rapidly achieve higher measles vaccination and ITN coverage is by distribution through mass campaigns. Vaccination campaigns, along with routine services, have been instrumental to reduce measles mortality. To accelerate reduction in overall child mortality, the WHO and UNICEF Global Immunization Vision and Strategy for 2006–2015 (World Health Organization 2005), a framework to fight vaccine-preventable diseases, includes a strategy to integrate immunization activities with other child survival interventions. ITN distribution has been integrated with measles vaccination campaigns in several countries in Africa, and studies have shown that this distribution method has led to substantial increases in ITN coverage (Grabowsky et al. 2005a,b; Centers for Disease Control and Prevention 2005; Mueller et al. 2008). It has also been suggested that integrating ITN distribution might increase vaccination coverage, especially among the poor, by providing added incentive for parents to bring their children; however, this effect has not been formally evaluated (Wolkon et al. 2010).
Madagascar is an island nation off the southeastern coast of Africa, with a population of approximately 19 million, 71% of whom live in rural areas (Central Intelligence Agency 2009). Approximately 90% of the population live in malaria-endemic areas (Kulkarni et al. 2010), and the disease is responsible for killing more children under the age of 5 years than any other cause; out of the 60 000 child deaths every year in Madagascar, malaria accounts for nearly one third (United Nations Children’s Fund 2006).
In October 2007, the Government of Madagascar, in conjunction with international partners, launched a nationwide integrated campaign targeting 3.1 million children <5 years of age for measles vaccination (age 9–59 months), and for mebendazole (age 9–59 months) and vitamin A (age 6–59 months) administration. In the 59 malaria-endemic districts where ITN distribution had not previously been conducted, long-lasting ITNs were also distributed to children aged 0–59 months. The Ministry of Health and the Malagasy Red Cross distributed more than 1.5 million ITNs together with measles vaccinations at fixed and temporary campaign sites, giving one ITN per child up to a maximum of two ITNs per household (Kulkarni et al. 2010). In this study, we evaluated the effect of integrating ITN distribution on measles vaccination coverage by comparing coverage in districts with and without ITN distribution, as well as the effect of ITN distribution on economic equity in measles vaccination.
Methods
Study design and data collection
In April 2009, 6 months after the campaign, we conducted a national, population-based, cross-sectional survey to estimate measles vaccination coverage, and ITN ownership and utilization. A stratified three-stage cluster sampling method was used to identify enumeration areas and households. Ten districts were chosen from each of three operational zones corresponding to 52 districts in the area without integrated ITN distribution, 33 districts in the area with integrated ITNs distributed by the Ministry of Health, and 26 districts in the area with integrated ITNs distributed by the Madagascar Red Cross. Districts were selected by probability proportional to size (PPS) sampling based on 2008 population projections from the 1993 census (Institut National de la Statistique 2003). In each selected district, six fokontany (the smallest administrative unit) were selected, plus alternates to replace any that were inaccessible, using PPS sampling. In each selected fokontany, all households were mapped using global positioning systems (GPS) receivers; 24 households and 6 alternates were randomly selected using PDA-based software (GPS Sample, CDC, Atlanta, USA) (Vanden Eng et al. 2007). All mothers or caretakers of children in the household were surveyed. This target sample size was calculated to detect at least a 7% difference in measles vaccination coverage between districts with and without ITN distribution, with 90% power, assuming 66% of households would have at least one eligible child, a design effect of two, and 80% coverage of measles vaccination during the campaign.
Trained teams conducted interviews using an electronic, standardized questionnaire programmed with Visual CE 10.2 software (Syware Inc., Cambridge, MA, USA), and entered into personal digital assistants (PDAs; Asus MyPal 696). Interviewers collected information on household socio-demographic characteristics, household ITN ownership and usage by household members and receipt of an ITN during the campaign. ITN ownership and type (i.e., ITN distributed through the integrated campaign, ITN distributed through other channels) were determined by inspection at the time of the interview. Information was collected on receipt of measles vaccination during the campaign and through routine vaccination services for all children who were 9–59 months of age at the time of the campaign. Measles vaccination histories were obtained by parental recall for vaccination through routine vaccination services, and by inspection of the campaign card, if available, or parental recall if not available, for vaccination administered during the campaign. Respondents with unknown vaccination status were assumed to be unvaccinated.
Statistical analysis
Survey data from PDAs were entered into a Microsoft Access database; analyses were conducted in SAS version 9.2 (SAS Institute, Cary, NC, USA) and SUDAAN version 10 (Research Triangle Institute, North Carolina). Using loading factors from the 2004 Madagascar Demographic and Health Survey, household wealth was determined by a composite score obtained from a principal components model using variables (e.g., whether the house had electricity, whether the family owned a bike or a radio, type of floor material and toilet, source of drinking water) from the survey. Wealth scores were partitioned to create wealth quintiles.
Population-level coverage estimates were calculated accounting for the sampling design (stratification and district-level clustering) and appropriate sampling weights. A finite population correction factor was incorporated to account for large sampling fractions of districts within each stratum. Taylor linearization method was used for variance estimation. Equity ratios for the national survey were estimated via relative risks comparing coverage in the poorest quintile and the wealthiest quintile; ratios <1 imply lower coverage among poorer households compared with wealthier households.
All surveyed districts were included for the national coverage estimates. The remaining analyses excluded the 20 Central Highland districts with low malaria transmission, and for which ITN distribution was not part of the national malaria control strategy.
Districts were not randomly assigned to receive ITNs during the campaign, but rather ITN distribution was purposively conducted in malaria-endemic areas that had not previously been targeted for net distribution. Because of the non-randomization of the intervention, an assessment of baseline demographics of malaria-endemic areas with ITN distribution (ITN districts) and without ITN distribution (non-ITN districts) was completed. To address imbalances found in observed characteristics between samples in areas with and without ITN distribution, we used the method of propensity score matching to create a subset of children within the two areas that were more balanced (Rosenbaum & Rubin 1985). In brief, logistic regression was used to estimate the predicted probability (i.e. propensity score) of a child being from an ITN district, including household- and child-level demographic factors as independent variables. A 1-to-1 matching algorithm was used to match children in ITN districts to similar children in non-ITN districts based on the propensity score (Parsons 2011). We fit several models with different sets of independent variables, creating unique matched subsets for each. Balance between the ITN district and non-ITN district subsets from each model was assessed by comparing the distributions of (1) each individual independent variable, regardless of whether it was included in the model or not, (2) histograms of propensity scores estimated from the model, and (3) histograms of propensity scores estimated from a refitted model based on the observations in the matched subsets. Based on these assessments, a final model was selected.
To evaluate whether the ITN integrated campaign resulted in higher measles vaccination coverage compared to the non-ITN integrated campaign, we estimated the relative risk (RR) of vaccination in ITN districts and non-ITN districts, and calculated 95% confidence intervals (CI) using log-binomial models in the propensity-matched subset. Equity ratios were estimated from the same model. For this subset analysis, standard generalized linear model methods (GLM) were used. Model building followed a forward-selection approach, assessing whether household wealth, highest level of education among women in household, previous net ownership, and child’s sex, age or previous routine measles vaccine dose impacted the association of the intervention on coverage. Two-way interactions were assessed independently.
Role of funding source
The funding sources had no involvement in the study design, data collection, data analysis, data interpretation, writing of the report, or in the decisions to submit the paper for publication. The corresponding author had full access to all the study data and had final responsibility for the decision to submit for publication.
Results
Vaccination coverage estimates
Interviewers collected data on 3158 children who were 9–59 months of age at the time of the campaign from 4302 households with an average of 1.7 children 9–59 months of age per household in 180 fokontany. Approximately 90% of households were in rural fokontany; due to flooding, 16 of the 180 selected fokontany were inaccessible and were replaced by previously selected alternates. Measles vaccination during the campaign was determined by parental recall and campaign card for 57.1% and 42.9% of children, respectively. Unknown measles vaccination status was reported for 1.4% of children through the campaign, and 2.7% through routine vaccination services.
Nationwide estimated measles vaccination coverage among children was 73.1% (95% CI 68.1–77.5) through routine services and 66.9% (95% CI 63.0–70.7) during the 2007 campaign (Table 1). Nationwide, 15.4% (95% CI 10.1–22.7) of eligible children received their first dose of measles vaccine during the campaign, 88.4% (95% CI 86.2–90.3) received one dose of measles vaccination either through routine or the campaign, and 51.6% (95% CI 43.4–59.7) were reported to have received both doses.
| Measles vaccination | Malaria-endemic areas | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Nationwide (n = 3158) | ITN districts† (n = 2233) | Non-ITN districts (n = 496) | |||||||
| n | % | 95% CI | n | % | 95% CI | n | % | 95% CI | |
| Routine | |||||||||
| Yes | 2350 | 73.1 | 65.0–79.9 | 1630 | 72.3 | 68.1–76.1 | 343 | 66.1 | 46.8–81.2 |
| Campaign | |||||||||
| Yes | 2233 | 66.9 | 61.6–71.9 | 1629 | 71.0 | 65.7–75.8 | 291 | 61.5 | 53.6–68.9 |
| Routine or campaign | |||||||||
| Yes | 2804 | 88.4 | 85.0–91.2 | 1981 | 88.6 | 83.8–92.1 | 421 | 86.1 | 81.1–89.8 |
- *Weighted estimates accounting for clustering.
- †Children residing in districts where insecticide treated nets were distributed (ITN districts) and were not distributed (non-ITN districts).
For the poorest to the wealthiest quintiles, respectively, nationwide routine measles vaccination coverage was 68.7% (95% CI 59.2–76.9), 66.3% (95% CI 53.8–77.0), 70.2% (95% CI 54.7–82.1), 80.2% (95% CI 73.1–85.9), and 79.7% (95% CI 74.9–83.8). The equity ratio (poorest vs. wealthiest quintiles) was 0.86 (95% CI 0.75–1.00). Nationwide measles vaccination coverage during the campaign was 65.5% (95% CI 55.8–74.0), 61.4% (95% CI 53.5–68.8), 71.5% (95% CI 65.3–77.0), 70.7% (95% CI 63.3–77.1), and 63.6% (95% CI 53.9–72.3), respectively. For the poorest quintile to the wealthiest quintile; the equity ratio was 1.03 (95% CI 0.89–1.19).
Evaluation of the integrated campaign: ITN districts and non-ITN districts
Children from ITN districts and non-ITN districts had similar age and sex distributions; however, children in non-ITN districts lived in households that were wealthier, had a higher level of education among women, and were more likely to own at least one non-campaign ITN (Table 2). The distribution of the propensity scores, estimated from the final logistic model that included household wealth score, type of flooring, radio ownership, highest level of education among women, number of children aged 9–59 months in the household, prior net ownership, and child’s sex, age and routine vaccination status as independent variables, shows the imbalance that existed in the samples taken from ITN districts and non-ITN districts (Figure 1: A vs. B). Table 2 shows the balance achieved for the household and individual child characteristics through propensity score matching. Propensity score matching resulted in nearly identical distributions of propensity scores in matched subsets of 315 children from ITN districts and non-ITN districts (Figure 1: C vs. D) thus reducing potential bias in the comparison of coverage in the ITN and non-ITN districts.
| Variable | Malaria-endemic areas | Propensity score subset | ||||||
|---|---|---|---|---|---|---|---|---|
| ITN districts (n = 2233) | Non-ITN districts (n = 496) | ITN districts (n = 315) | Non-ITN districts (n = 315) | |||||
| n | % | n | % | n | % | n | % | |
| Age in months | ||||||||
| 9–11 | 226 | 10.1 | 31 | 6.3 | 30 | 9.5 | 24 | 7.6 |
| 12–23 | 522 | 23.4 | 122 | 24.6 | 82 | 26.0 | 74 | 23.5 |
| 24–35 | 463 | 20.7 | 102 | 20.6 | 56 | 17.8 | 69 | 21.9 |
| 36–47 | 536 | 24.0 | 122 | 24.6 | 74 | 23.5 | 73 | 23.2 |
| 48–59 | 486 | 21.8 | 119 | 24.0 | 73 | 23.2 | 75 | 23.8 |
| Sex | ||||||||
| Male | 1079 | 48.3 | 255 | 51.4 | 173 | 54.9 | 168 | 53.3 |
| Wealth quintile | ||||||||
| 1 (poorest) | 552 | 24.7 | 18 | 3.6 | 33 | 10.5 | 18 | 5.7 |
| 2 | 612 | 27.4 | 140 | 28.2 | 90 | 28.6 | 99 | 31.4 |
| 3 | 444 | 19.9 | 128 | 25.8 | 68 | 21.2 | 82 | 26.0 |
| 4 | 377 | 16.9 | 104 | 21.0 | 60 | 19.1 | 62 | 19.7 |
| 5 (wealthiest) | 248 | 11.1 | 106 | 21.4 | 64 | 20.3 | 54 | 17.1 |
| Highest level of education among household women‡ | ||||||||
| None | 694 | 33.5 | 85 | 18.3 | 61 | 21.0 | 66 | 22.8 |
| Primary | 973 | 47.0 | 265 | 57.1 | 165 | 56.7 | 164 | 56.6 |
| Secondary or higher | 403 | 19.5 | 114 | 24.6 | 65 | 22.3 | 60 | 20.7 |
| ITN household ownership (≥1) | ||||||||
| Non-campaign ITN | ||||||||
| Yes | 645 | 28.9 | 353 | 71.2 | 183 | 58.1 | 196 | 62.2 |
| Routine measles vaccination | ||||||||
| Yes | 1630 | 73.0 | 343 | 69.2 | 226 | 71.8 | 221 | 70.2 |
| Number of household children eligible for measles vaccination during the 2007 campaign | ||||||||
| 1 | 1079 | 48.3 | 291 | 58.7 | 189 | 60.0 | 174 | 55.2 |
| 2 | 850 | 38.1 | 148 | 29.8 | 95 | 30.2 | 101 | 32.1 |
| 3 | 231 | 10.3 | 45 | 9.1 | 22 | 7.0 | 33 | 10.5 |
| 4 | 73 | 3.3 | 12 | 2.4 | 9 | 2.9 | 7 | 2.2 |
- *Observed sample, unweighted proportions.
- †Children residing in districts where insecticide treated nets were distributed (ITN districts) and were not distributed (non-ITN districts).
- ‡7% had missing value.
Distribution of propensity scores among all children (panels a and b) and the propensity score subset of children (panels c and d) living in districts with insecticide treated net distribution (ITN districts) and districts without ITN distribution (non-ITN districts). The propensity score model included the following variables: household’s wealth score, type of flooring in the household, owning a radio, highest level of education among women in household, number of children in the household aged 9–59 months, prior ITN ownership, and child’s sex, age and routine vaccination status.
The final model comparing campaign vaccination coverage in the ITN and non-ITN districts used the propensity score subsets and included the potential effect modifiers of wealth quintile, previous routine measles vaccination and prior net ownership. Model results showed campaign vaccination coverage was higher in ITN districts (70.8%) compared with non-ITN districts (59.1%) when averaging over the wealth quintiles (RR = 1.3, 95% CI 1.1–1.6) (Table 3). Among children from the poorest households, children in ITN districts were twice as likely to have received measles vaccination during the campaign compared with those in non-ITN districts (wealth Q1: RR = 2.4, 95% CI 1.2–4.8; wealth Q2: RR = 1.8, 95% CI 1.4–2.4). No significant differences were found in the wealthiest three quintiles (wealth Q3: RR = 1.0, 95% CI 0.8–1.2; wealth Q4: RR = 0.9, 95% CI 0.7–1.2; wealth Q5: RR = 1.0, 95% CI 0.9–1.2). Improved equity for measles vaccination during the campaign was achieved in ITN districts (equity ratio = 1.0, 95% CI 0.8–1.3) compared to non-ITN districts (equity ratio = 0.4, 95% CI 0.2–0.8).
| ITN districts (n = 315) | Non-ITN districts (n = 315) | Vaccinated during Campaign ITN vs. non-ITN districts | ||||
|---|---|---|---|---|---|---|
| n | % | n | % | RR† | 95% CI | |
| Overall | 223 | 70.8 | 186 | 59.1 | 1.3‡ | 1.1–1.6 |
| Wealth quintile | ||||||
| 1 (poorest) | 24 | 72.7 | 6 | 33.3 | 2.4 | 1.2–4.8 |
| 2 | 65 | 72.2 | 38 | 38.4 | 1.8 | 1.4–2.4 |
| 3 | 45 | 66.2 | 59 | 72.0 | 1.0 | 0.8–1.2 |
| 4 | 37 | 61.7 | 42 | 67.7 | 0.9 | 0.7–1.2 |
| 5 (wealthiest) | 52 | 81.3 | 41 | 75.9 | 1.0 | 0.9–1.2 |
- *Insecticide-treated net.
- †Relative risk (RR) estimated from a log-binomial model including the main effects for ITN, wealth quintile, routine measles vaccination, prior ITN ownership and the interaction ITN × wealth quintile.
- ‡Estimated RR averaged over 5 equal wealth quintiles.
Discussion
Both measles vaccination and ITN coverage increased nationwide after the integrated campaign. Distribution of ITNs along with measles vaccination during the campaign led to significantly higher and more equitable vaccination coverage and appeared to be an effective incentive for children to receive measles vaccine among poorer households. These results suggest that the integrated campaign strengthened both measles vaccination and ITN distribution efforts, especially among the poor, and created synergy with implementation of these two important child survival interventions in Africa.
The results of this study should be considered in light of limitations. First, selection of districts for ITN distribution during the campaign was not random, but was based on malaria endemicity and previous ITN distribution. We attempted to control for potential non-randomization bias through propensity score matching using multiple background characteristics; however, some bias may have remained due to unmatched factors. Further evaluations of the effect of integrating ITN distribution and vaccination campaigns are needed; however, careful consideration should be given to the target population and to the feasibility of implementing a large-scale, cluster-randomized controlled study. Second, vaccination status was based primarily on parental recall, and may have been inaccurate; however, the survey was conducted within 6 months after the campaign and there was no reason to expect that the two subsets used for comparison remembered receipt of vaccination differently. Finally, results of this study might not be representative of all of Madagascar, as 9% of selected fokontany were inaccessible; however, we attempted to minimize any bias by replacing inaccessible fokontany with previously-selected alternates from the same districts.
During 1990–2008, substantial progress towards Millennium Development Goal (MDG) 4 has been made: child mortality fell by 28% globally and by 22% in sub-Saharan Africa (Black et al. 2010; You et al. 2010). Because of high measles mortality among children in low-income countries, measles vaccination coverage at one year of age is one of three indicators used to measure progress toward MDG 4 (United Nations 2011). During 2000–2008, coverage with at least one dose of measles vaccine in World Health Organization Africa Region countries increased by one-fifth from 56% to 67% (World Health Organization 2011), and estimated measles deaths dropped by 92% from 371 000 to 28 000 (Centers for Disease Control and Prevention 2009). Progress towards achieving MDG 6 has also been made: from 2000 to 2006, use of ITNs among children aged <5 years increased 10-fold, from approximately 2% to 20% (United Nations 2011). In several countries with high coverage with malaria interventions, including ITNs, indoor residual spraying, and artemisinin-based combination therapies, severe malaria cases and deaths among children in health facilities have declined by ≥50% (United Nations 2011). The efficacy of these interventions is well documented; however, it should be noted that one study, recently published, described the potential impact of increasing resistance to pyrethroid insecticide in ITNs among A gambiae mosquitos in Senegal (Trape et al. 2011). Although this report raised some concerns, the positive public health benefit of ITNs is established and ITN use remains the primary recommended means of reducing malaria (Roll Back Malaria Partnership 2005, 2008).
Despite the progress toward achieving the MDGs, an estimated 8.8 million children <5 years of age died in 2008, including 4.4 million deaths in sub-Saharan Africa. To further reduce child mortality and to reduce persistent health disparities between the rich and the poor, strategies are needed to improve coverage with measles vaccination and ITNs. The success of mass vaccination campaigns, as a service delivery model, in reaching high coverage with measles vaccination is well documented (Biellik et al. 2002; Otten et al. 2005). Mass campaigns are cost-effective (Uzicanin et al. 2004), and can improve equity within populations (Vijayaraghavan et al. 2007; Wolkon et al. 2010). Because of these strengths, other health programs have sought to use measles vaccination campaigns to deliver services such as vitamin A administration, albendazole administration, and ITN distribution (Centers for Disease Control and Prevention 2005). However, integration of additional services adds complexities to campaign planning and implementation (Centers for Disease Control and Prevention 2005). For example, ITN procurement, shipping, storage, and transport can be challenging during integrated campaigns (Grabowsky et al. 2005a), and delays or stock-outs may undermine efforts to maximize campaign attendance (Grabowsky et al. 2005a; Centers for Disease Control and Prevention 2005), thus negatively impacting vaccination coverage. On the other hand, the added incentive created by free distribution of ITNs that carry a monetary value in the marketplace may increase campaign attendance (Grabowsky et al. 2005b; Mueller et al. 2008), especially among the poorest children as demonstrated by our results.
Equitable distribution of public health services is essential for protecting the basic human right of access to health services (Backman et al. 2008; Pillay 2008). Previous studies have shown that integrating the distribution of ITNs with measles vaccination campaigns can increase both ownership and usage of ITNs and improve equity in ITN ownership (Grabowsky et al. 2005a, 2007; Thwing et al. 2008; Wolkon et al. 2010); our findings show improved equity in measles vaccination coverage, thus providing protection for the most vulnerable and difficult to reach children. Despite the inherent operational challenges, implementation of this strategy in other countries might accelerate progress toward MDG 4 and MDG 6 and control of both measles and malaria, two major causes of childhood morbidity and mortality in the world today.
Acknowledgements
The authors are grateful to the families who participated in this survey and to teams who worked under challenging field conditions to complete this survey. We would like to thank the many international partners who contributed to the implementation of the integrated campaign and its evaluation, including the Madagascar Ministry of Health, Family Planning and Social Protection, Madagascar Red Cross, CRESAN, Malaria No More, UNICEF, World Health Organization, Canadian International Development Agency, USAID, Canadian Red Cross, Institut National de la Statistique, Institut National de Sante Publique et Communautiare. Special thanks go to Dr. Izaka Rabeson, Dr. Bary Rakototiana and Mr. Fanja Ratsimbazafy of the Madagascar Red Cross for their logistical and administrative support and supervision in the field. In addition, we are grateful to Dr. Elizabeth Zell at the US Centers for Disease Control and Prevention for her valuable input on propensity score methods. The work was funded by the Canadian International Development Agency, United States Agency for International Development, Canadian Red Cross, and United Nations Children’s Fund. The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention.
