Background
Child survival came to the forefront of the global health agenda with the 2003 series in The Lancet [
1‐
5]. The opening report stated that an estimated 10.8 million children aged under-five were dying globally in 2000. Approximately 90% of these deaths were occurring in 42 countries. The major causes of mortality were identified as diarrhoea and pneumonia, whilst neonatal disorders, malaria and HIV contributed substantially in certain regions [
1]. An estimated 63% of these under-five deaths could be prevented if current knowledge on available and feasible interventions is translated into effective action [
2]. One of the targets of the United Nations' Millennium Development Goals (MDGs) is to reduce the under-five mortality rate by two-thirds between 1990 and 2015 (MDG4). UNICEF, WHO and other experts agreed on a minimum set of key indicators for monitoring progress to achieve MDG4. These include oral rehydration therapy (ORT) and continued feeding for diarrhoea, the use of insecticide treated mosquito nets (ITNs) and antimalarial treatment for under-fives, antibiotic treatment for pneumonia, immunization coverage (tetanus, measles and DTP3), and timely initiation of breastfeeding and exclusive breast feeding until 6 months [
6].
There is evidence to suggest that participatory health education interventions and/or community-based primary health care in remote regions can improve child survival [
7,
8]. These community-based interventions target health knowledge and promote appropriate treatment-seeking and preventative behaviours. In Nepal, an intervention using women's health groups led to a reduction in neonatal mortality of nearly 30%.These groups met monthly to identify problems and implement and evaluate strategies to tackle the identified problems [
9]. A similar approach saw a significant decrease in perinatal mortality in Bolivia [
10,
11]. In India, home-based neonatal care and management of sepsis by trained community health workers led to a 62% decline in neonatal mortality when compared to a control region that surrounded the project area. Moreover, fatality from sepsis declined from 16.6% to 2.8% [
12]. The evidence suggests that an intensive community health education programme in combination with mentoring of community health workers and improved availability of services and materials at the community level may lead to major, cost-effective declines in child mortality.
The most recent data ranks Guinea Bissau 5
th from the bottom globally. The neonatal mortality rate was 47 per 1000 live births in 2004 and the country had an under-five mortality rate of 198 per 1000 live births and an infant mortality rate of 118 per 1000 live births in 2007 [
13]. There is very little information on the causes of child deaths in Guinea Bissau, as the country was not included in the epidemiological profiles supplied in the Child Survival series. However, nearby countries were included in Profile 2 and diarrhoea, malaria, pneumonia and neonatal disorders were identified as the leading causes of child mortality [
1].
EPICS (Enabling Parents to Increase Child Survival) is a cluster randomised trial, which is currently running in rural areas of southern Guinea Bissau [
14]. It aims to evaluate whether an intervention package can generate a rapid and cost-effective reduction in under-five child mortality. The package includes health education through health clubs and house-to-house visits. It also includes intensive training and supervision of village health workers and community-based birth attendants, improved outreach services and enhanced institutional delivery services. A maternal birth history survey conducted in the region before the trial started in 2007 found an under-five child mortality rate of 154 per 1000 live births, as reported by women aged 12 to 49 for their 2002 to 2007 pregnancy history. This is somewhat lower than nationally reported [
13].
The aims of this study were to understand levels of knowledge on child health as well as the treatment-seeking and preventative behaviours of the inhabitants of the trial area. The study was conducted prior to the start of the trial in order to develop an appropriate and effective health education component. The study also aimed to assess the effect of gender and ethnicity on knowledge and behaviour.
This article reports on the knowledge and reported practices of men and women on maternal and child health in two districts of southern Guinea Bissau. It also reports the effect of gender and ethnicity of knowledge and behaviour.
Methods
Study setting and participants
The study was conducted in the two rural districts of southern Guinea Bissau (Quinara and Tombali) in which the EPICS trial is being implemented. Villages in the region were considered eligible for the trial if they were believed to have a population of approximately 300 inhabitants. The villages were mapped and enumerated during the maternal birth history baseline survey. Clusters were formed from one or more villages in order to create an average population of 350 per cluster. This process resulted in 241 villages being organised into 146 clusters for inclusion in the trial [
14]. This baseline study of health knowledge and practice was conducted in 31 of these villages after the clusters had been identified but before they were randomised into control and intervention arms. The villages were selected on the basis of their accessibility (in the rainy season, when this study was conducted, some of the villages are inaccessible). The selected villages included those that had majority Fula (13), Beafada (10) and Nalu (5) populations. These are ethnic groups that normally identify themselves as Muslim. Two villages had Susu populations and one had Balanta population as majority.
In each selected village, 10 households out of those that had already been mapped were randomly selected for this study. Within each household, one woman who was a primary carer of a child aged five or under and one man who lived in the same household as the woman (most commonly her husband) were selected for interview. If a visited household had more than one eligible woman and/or man, the ones who were interviewed were randomly selected.
Data collection and instruments
A fully structured questionnaire (Additional file
1) was developed and interviews were carried out by trained field workers who were supervised. The questionnaire -- which was in part based on questions from DHS surveys - collected information on knowledge and reported practices on the key indicators affecting child survival [
3]. It was arranged into seven sections on household characteristics, respondent characteristics, health knowledge, pregnancy and delivery, vaccination status, treatment-seeking behaviour and household observation. Section one was completed with the woman and man together. Sections two, three and parts of six were completed with both the woman and man, interviewed separately. Sections four, five, parts of six and seven were completed only with the woman. Sections one and two had questions with only one possible answer. The other sections had two types of question: those for which only one answer could be given and those for which more than one answer was possible.
The questionnaire was piloted in two different villages in southern Guinea Bissau and revisions were made before the survey was conducted. These villages were excluded from the survey. A team of six field workers -- recruited from Bissau - and one field supervisor were provided with intensive classroom and field-based training as well as a detailed manual of survey guidelines. Field workers conducted the survey in Kriol or, where necessary, directly or through translation in appropriate local languages.
Data analysis
Characteristics of the households and of the interviewed women and men were tabulated. These included background characteristics of the households and demographic characteristics of the respondents by gender. Descriptive statistics for continuous variables included the mean, standard deviation, median, and range and for categorical variables the numbers and percentages. The number of correct answers given to the health knowledge and practice questions and their percentage distribution were tabulated by items and by gender (as appropriate).
An overall health knowledge score was derived by summing the total number of correct responses given to the fifteen health knowledge questions in section three of the questionnaire. Questions with more than one possible correct answer were weighted by the inverse of the total number of correct answers so that no single question contributed more than one point out of fifteen to the score. Factors affecting health knowledge scores were investigated using simple linear regression models. Gender was used as a single binary predictor. Education was categorised into primary (grade 1-6), secondary (grade 7-9), higher (grade 10-12) or Koranic education. Ethnicity was included in the models as Balanta, Beafada, Fula, Nalu, Mandinka, Susu and Bijagos, or other. Age and an indicator of household crowdedness (the number of people per room within the house) were used as continuous variables. Robust standard errors were used to allow for clustering within villages. The final multiple linear regression includes gender and ethnicity and other factors which either had a statistically significant effect on obtained health knowledge score themselves or confounded the effect of gender and/or ethnicity.
Ethical considerations
This study was approved by the ethics committees of 'The Centre for Coordination of the Research of the Ministry Of Health' in Guinea Bissau (NCP-021/2007) and the London School of Hygiene and Tropical Medicine (5173).
Discussion
The study found low levels of appropriate knowledge on child health, some inappropriate practices and generally low vaccination coverage. The study showed that health knowledge scores improve significantly amongst those who have accessed education, in line with other studies [
15,
16]. Regression analysis revealed that differences in health knowledge between women and men become insignificant once age and education are accounted for.
Most of the results presented in this paper are in line with the UNICEF findings on Guinea Bissau in their 2008 report, which tracks the key indicators for monitoring progress to achieve MDG 4 [
17]. Any differences might be explained by the fact that part of this study focused on knowledge, whereas the UNICEF report details reported practice. Furthermore, WHO and UNICEF estimate immunisation coverage of 89% for BCG, 63% for 3 doses of DPT, 64% for 3 doses of polio (OPV), and 76% for measles in 2007 [
18]. This is very close to the findings of this study of 88%, 66%, 62% and 77% respectively for the same year.
This study has some limitations. First, the findings of this survey on levels of health knowledge and treatment-seeking and preventative behaviour are only representative of the area selected for the EPICS trial rather than the population of southern Guinea Bissau as a whole. Second, the findings could be biased due to the constraints of conducting research in the difficult conditions of the rainy season. Villages were selected for this survey on the basis of their accessibility, rather than randomly. This resulted in higher numbers of Fula (43%), lower numbers of Beafada (31%) and very low numbers of Balanta (5%) respondents compared to the representation of these groups within the population, based on the enumeration data for the EPICS trial on 7876 women (Beafada (35%), Fula (17%) and Balanta (21%)). This is probably due to the fact that the clustered settlements in which the Beafada and Fula most commonly live are often found in more accessible on-road locations, whereas the scattered settlements in which the Balanta most commonly live are often found further away from the main roads.
Despite its limitations, the method employed in this study has some useful implications for future research. Very few studies of health knowledge and behaviour bring together the range of issues covered in this survey. The majority of studies were conducted on mothers or women of reproductive age on single themes such as the prevention and management or treatment of diarrhoea [
19,
20], malaria [
21,
22], and pneumonia [
23,
24], as well as on knowledge and practice surrounding antenatal care, delivery [
25,
26] and breastfeeding [
27,
28]. There are a number of studies on health knowledge -- particularly on malaria - which were conducted on both women and men (parents, caregivers, or community members in general) [
29‐
31]. Of the studies mentioned above, many were conducted at the facility level and/or in urban settings and very few studies compared female and male knowledge and practice on a range of issues related to child health at the household level in rural settings. An exception is a publication, which examined community effects such as the proportion of literate adults and the presence of a market, on health knowledge in Ghana [
15]. As integrated approaches to the reduction of child mortality at the community level become more widespread, it becomes critical to develop tools that can capture a range of knowledge and practice of women and men at the household level in rural settings in order to plan appropriate and effective interventions. The authors believe the survey conducted in this study is a useful contribution to this.
Furthermore, most studies of health knowledge that have been conducted in rural areas of developing countries have reported only descriptive statistics, the numbers and/or percentages of respondents giving correct answers or all the possible correct answers for each individual health knowledge related question, or they derived a simple scoring system which increases by one for each correct answer [
32,
33,
24]. In this study, an overall health knowledge score, which is weighted by the inverse of the possible correct answers, has been developed. The authors have used regression analyses, which take into account the clustering of the sample in the villages by using robust standard error estimates, to assess the association between factors and overall health knowledge. As far as the authors are aware, only one group has allowed for hierarchical structure in their analyses [
15]. This approach creates new possibilities for measuring changes in health knowledge on multiple issues over time.
Finally, one of the aims of this study was to gather information to inform the development of an appropriate and effective health education programme in southern Guinea Bissau. The study provided critical information for programme planning both in the short and long term. In the short term, targeted programmes have been developed that address key gaps in health knowledge and practice, as outlined by the findings of this survey. These programmes primarily target women as the primary carers of children aged under-five, but they also include men who, as this survey shows, are frequently referred to by their wives when a child is sick. Specifically, the health club methodology [
34] has been adapted to address health knowledge and practice in relation to child survival. Health clubs are community associations that hold weekly meetings to discuss health issues in a highly structured participatory framework. This survey provided critical information to develop the content of the health clubs for the intervention (EPICS trial) in southern Guinea Bissau. In the long term, programmes must be developed that address the chronic lack of access to education in the region, particularly amongst women.
Conclusion
Prior to planning community-based health education interventions, it is important to conduct properly designed surveys to assess levels of knowledge on child health, treatment-seeking and preventative practices as well as vaccination coverage.
This study found low levels of appropriate health knowledge on diarrhoea, malaria and pneumonia as well as on breastfeeding and risk signs during pregnancy. It also found that the majority of births occurred at home without the presence of a skilled birth attendant and that poor practices existed in relation to the care of the umbilical cord and immediate breastfeeding. The results also showed inappropriate levels of vaccination coverage. Therefore, health education activities that target these areas of knowledge and practice should be an integral part of an intervention package to improve child survival in rural Guinea Bissau.
Although child health education activities often target women only, our study in rural Guinea Bissau showed it could be crucial to involve men in health education activities to improve child survival.
Finally, the derivation of a weighted overall health knowledge score and its use in robust regression analysis is addressing an important methodological issue in assessing health knowledge on multiple topics.
Acknowledgements
The authors would like to thank the following for their contributions: Kalie Sassoon, Albino dos Santos and the field team for acquisition of data; Mark Fisher for database design; Ila Fazzio, Jose Carlos de Sa and Polly Walker for advice on study design; Adil Shah for logistical support; and Chitra Jayanty for advice and support.
All individuals mentioned above were funded by Effective Intervention.
This study, the authors, and the preparation of the manuscript were funded by Effective Intervention.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
RK contributed to the conception and design of the study, the acquisition of data, the interpretation of the data, and the drafting of the manuscript. VM contributed to the conception and design of the study, the analysis and interpretation of the data, and the drafting of the manuscript. PB contributed to the conception and design of the study and the revision of the manuscript. All authors read and approved the final manuscript.