In developing countries, traditional healers often have the role of being the primary health care providers for their communities 123. In addition to traditional birth attendants (TBAs), there are distinct groupings of traditional healers that provide primary health care in communities, in different forms based on their skill level, their accessibility, and whether they underwent lengthy apprenticeships or a spiritual "calling" to their role 456.

Although many traditional healers are herbalists, this is not the only way traditional health care is practiced. Some call upon the ancestral spirits or perform exorcism to treat an illness, yet the herbalist may also incorporate this spiritual aspect in diagnosing the patient's illness 357. Faith healers may utilize prayer, touch, and ointments in their healing rituals. There are also healers who combine Islamic medicine, and will invoke verses of the Koran and/or use astrology in the healing process 3.

The services of a traditional healer may be sought in cases where the illness is considered 'native' in origin. However, if the illness is considered to have a foreign origin, a Western medical provider may be consulted, out of the belief that they will be more familiar and have better treatment. The treatment is determined as specifically relating to the source of the illness, whether worldly or other-worldly. The use of traditional healers is widespread and accepted in most Haitian communities. Twenty-three percent of mothers go to traditional healers when their children are sick 8.

A Presidential decree of April 4, 2003 recognized Vodou as a bonafide religion in Haiti. With regard to the decree, Vodou chiefs, temple officials, officials at any sacred site, as well as all Vodou organizations or associations were empowered to file a request for recognition by the Ministry of Culture and Religious Affairs. It is possible for people in Haiti to belong to a mission religion such as Catholicism or Protestant Christianity but still practice Vodou 910.

According to the World Health Organization (WHO), the two public health interventions that have had the greatest impact of the world's child health are clean water and vaccines. In light of this fact the importance of encouraging vaccination in Haiti is obvious benefit. Vaccinations against tuberculosis, diphtheria, pertussis, tetanus, measles, and polio have dramatically reduced the burden of death and disease from infectious diseases 1112. However, vaccination services continue to be under-utilized, especially in developing nations. Almost two million children still die each year from vaccine-preventable infectious diseases 13. In Haiti, more than 138,000 children under five years of age die of preventable diseases annually; 60% of those who survive may fail to develop adequately 14. The major causes of mortality among children under five years old are diarrhea, malnutrition, measles and malaria. Furthermore, because measles enhances deficiency in vitamin A, it continues to be an important cause of blindness 15. Only 30% of Haitian children 12–23 month old are fully vaccinated. There appears to be no sex or urban versus rural differences in vaccination prevalence in the country 8.

Over the years, the World Health Assembly has adopted a number of resolutions encouraging the study of the potential usefulness of traditional medicine including evaluation of practices and examination of safety and efficacy of remedies used. The WHO also advocates for the education and information dissemination of the public about proven traditional health practices 161718. Globally, traditional healers have been reported to offer treatments for hypertension, 19 cancer, 20 AIDS, 21 tuberculosis, 2223 malaria, 32425 sexually transmitted infections, 26 epilepsy, 27 and infertility 28. However, there is paucity of studies on the role of traditional healers in vaccination. The purpose of this study was to examine the association between the use of traditional healers by mothers and child vaccination.

Of the 720 eligible mothers, 691 (96%) completed the survey. Table 1 presents selected characteristics of the study population (mean age 28.9 years, standard deviation 10.1 years). Most of the respondents were in the age group 25–34 years old (42.6%), and had elementary education (52.5%).

Those who were affiliated to Vodou were 32.1% of the sample. Nine in ten children (93.6%) had BCG (Bacille Calmette-Guérin), the vaccination against tuberculosis and one in two had measles vaccination (51.3%). Only 47% of children five years of age or younger were fully vaccinated. About one half of mothers reported using the services of traditional healers (48.3%).

In this study, we found that 47% of the subjects were fully vaccinated, which is a slightly higher than the national average of 43%. Vaccination coverage in Haiti is far below the 78% global vaccination coverage 30 and the 88% vaccination coverage of the Pan American Health Organization (PAHO) region 31.

The use of traditional healers was negatively associated with vaccination. This may be an indication that traditional healers do not support the use of the formal health sector which includes vaccination. Burnet and Buggaley 32 report that there is often mistrust and lack of understanding between traditional healers and the formal health sector, but traditional healers are more open for collaboration than formal health workers. Several studies have indicated that collaboration between traditional healers and the formal health care sector is possible, useful and crucial in order to bridge the treatment gap in developing countries 2533343536.

Vodou affiliation was associated with decreased vaccination rate, which may be an indication that Vodou practice may not be favorable to vaccinations. It is also possible that Vodou affiliation may just be an indicator of some unmeasured variable which may also be associated with non-vaccination of children.

The United Nations Children Fund (UNICEF) recommends that public health workers collaborate with religious leaders and traditional healers whose support can help reach out to members of the community who are reluctant to vaccinations 37. It is encouraging to note that a number of modern health care groups have already started working with Vodou priests in Haiti to help identify people who are HIV-positive 38. While this recommendation is appropriate, only a future evaluation will show or fail to show if collaboration with Vodou priests will improve the prevalence of children who will be fully vaccinated.

Our findings indicate that the proximity to a health care facility was positively associated with full vaccination. Child vaccinations are provided for free at the clinic in Pont-Sonde and at the nearest hospital. Previous studies have identified long distances from vaccination site as a risk factor for failure to be vaccinated 3940. Reichler et al 39 reported that in Egypt, being farther than 10 minutes by foot from the nearest National Immunization Day (NID) site was a risk factor for failing to receive NID vaccines. In Kenya, Ndirutu et al 40 found that vaccination coverage was reduced with every kilometer of distance from home to vaccination clinic. In Haiti it might be worthwhile to facilitate intermittent vaccination campaigns within the community which may reduce the distance from homes to vaccination sites.

In this study, maternal education was not associated with child vaccination. At least three-quarters of women in our sample had elementary education only or none. Unlike in other settings where women with low education or none may be worse off than others, we find that there was not a difference in vaccination status based on maternal education. This may have been that there was limited variability in resources (cognitive or material) that may have influenced decision to vaccinate e.g. the knowledge, attitudes and behaviors which influence decisions to vaccinate may have been similar and educated women did not have better financial resources to facilitate transport to clinic 4142.

This study had several limitations. Firstly, this was a cross-sectional study and we, therefore, cannot assume causality i.e. none of the factors associated (positively or negatively) with vaccination can be taken as have caused the vaccination. Secondly, some participants may have mis-reported their use of traditional healers. Although the use of traditional healers is widely accepted in Haiti, some medical professionals discourage or have negative attitudes towards the practice. This may have led some mothers to underreport their use of traditional health services; in such case, the reported odds ratios would be conservative values. Furthermore, we categorized Vodou affiliation mutually exclusive with other religions. To the extent that mothers affiliation to Vodou and other religions, our results may be biased. However, such bias may be that the measure of association (the OR) would tend to move towards the null (equality). This would mean that the effects that we have estimated are in fact under-estimates.

Among Haitian children, we found that mother's use of traditional healers was negatively associated with vaccinations. Findings from this study underscore the need to undertake education efforts towards traditional healers and to include them in vaccination efforts. Given the success other areas of modern health care have had in working with traditional healers, including vaccination programs in that collaboration could considerably increase vaccination coverage in Haiti. This approach may also be explored in other developing countries experiencing low vaccination coverage.

The authors declare that they have no competing interests.

ER participated in the study design, conducted data analysis and participated in the drafting of manuscript. MYP participated in study design, conducted data collected and participated in the drafting of manuscript. JJ participated in study design, interpretation of findings and drafting of manuscript. SS participated in the interpretation of findings and drafting of the manuscript. ASM led the manuscript drafting effort and participated in the interpretation of findings. All author approved the final draft of the manuscript.