Menu
Menu
Menu
Menu
Background: Guinea-Bissau is a small country in West Africa with a population of 1.7 million. The WHO and UNICEF reported an under-five child mortality of 203 per 1000, the 10thhighest amongst 192 countries. The aim of the trial is to assess whether an intervention package that includes community health promotion campaign and education through health clubs, intensive training and mentoring of village health workers to diagnose and provide first-line treatment for children’s diseases within the community, and improved outreach services can generate a rapid and cost-effective reduction in under-five child mortality in rural regions of Guinea-Bissau. Effective Intervention plans to expand the project to a much larger region if there is good evidence after two and a half years that the project is generating a cost-effective, sustainable reduction in child mortality. Methods/design. This trial is a cluster-randomised controlled trial involving 146 clusters. The trial will run for 2.5 years. The interventions will be introduced in two stages: seventy-three clusters will receive the interventions at the start of the project, and seventy-three control clusters will receive the interventions 2.5 years after the first clusters have received all interventions if the research shows that the interventions are effective. The impact of the interventions and cost-effectiveness will be measured during the first stage. The package of interventions includes a community health promotion campaign and education through health clubs, and intensive training and mentoring of village health workers to diagnose and provide first-line treatment for common children’s diseases within the community. It also includes improved outreach services to encourage provision of antenatal and post natal care and provide ongoing monitoring for village health workers. The primary outcome of the trial will be the proportion of children that die under 5 years of age during the trial. Secondary outcomes will include age at and cause of child deaths, neonatal mortality, infant mortality, maternal mortality, health knowledge, health seeking behaviour, morbidity and costs. Discussion. The trial will be run by research and service delivery teams that act independently, overseen by a trial steering committee. A data monitoring committee will be appointed to monitor the outcome and any adverse effects.
The study is a cluster-randomised trial, where communities will be randomised at the start of the trial to intervention and control clusters. Control clusters will receive the interventions approximately 2.5 years after the intervention clusters if the research shows that the interventions are effective. The primary outcome of the trial is the proportion of children that die under- five during the study period. Secondary outcomes will include: – neonatal and infant mortality rates, and the probability of death under the age of five calculated using life-table methodology – age at and cause of child deaths – treatment practices for sick children – percentage of mothers who use ORS to treat children with diarrhoea – percentage of children who received consultations with trained personnel when recently sick – mother’s or primary caregiver’s knowledge of childhood diseases and safe delivery – proportion of mothers with children under-five that have knowledge of symptoms of dangerous disease; – proportion of mothers with children under-five that have knowledge of treatment procedures and medicines usage for diarrhoea, fever, and cough with rapid breathing – proportion of women aged 15 to 49 with knowledge of: – key safe practices related to antenatal care – symptoms or factors that would imply a pregnant mother has a high risk of complications at delivery – need to sterilize instruments used to cut the umbilical cord – procedures to keep the umbilical cord sterile – symptoms of neonatal sepsis – recommended breast feeding practices after delivery – child morbidity (prevalence of fever, diarrhoea, respiratory tract infections) – maternal mortality rate per 100,000 pregnancies – age at and cause of maternal deaths – proportion of deliveries conducted at institutions – indicators of safe home birthing practices – proportion of deliveries where trained personnel were present – proportion of deliveries where instruments used to cut the umbilical cord were sterilized – proportion of deliveries where the belly button was treated with alcohol or antiseptic solution subsequent to cutting the umbilical cord – proportion of deliveries where measures to prevent hypothermia were taken – proportion of deliveries where breast-feeding was started within one hour after birth – cost effectiveness Clusters will be stratified according to major ethnic categories (Balanta, non-Balanta and mixed ethnicity) and to distance from a regional health centre or hospital (within walking distance or not). All 146 clusters will be randomly allocated, within these six strata, to either the intervention or control group. Table Table11 outlines the activities that each arm of the project receives during the first two and one half years. Trial activities by arms of the trial in the first 2.5 years At the start of the intervention, health clubs will be organised in each community. The number of health clubs will be chosen so that there are approximately 60 households per club. These clubs will meet regularly for two years and cover topics related to maternal and child health that are expected to impact on child mortality: • Recognition, treatment and prevention of malaria, pneumonia and diarrhoea • Safe pregnancy and delivery • Care for newborns • Hygiene and sanitation In addition to health clubs, communities will meet at the start of the interventions to select village health workers. These village health workers will be members of the health clubs, and will work closely with health clubs. The village health workers will be given intensive block training (3–5 days for each session) by registered nurses, “nurse trainers”, in the following areas: • Recognition, treatment and prevention of diarrhoea • Recognition, treatment and prevention of malaria • Recognition, treatment and prevention of pneumonia • Safe pregnancy and delivery • Care for the newborn • Nutrition and child growth monitoring The training for these workers will reflect the Ministry of Health’s policies and IMCI recommendations [10]. Village health workers will also attend and help manage their local health clubs, and they will help coordinate antenatal and vaccination service provision within the cluster. They will also assist with home deliveries and make regular visits to monitor neonates immediately after birth. They will also provide advice on treatment of sick children. Once they pass through the relevant stage in their training, they will be asked to coordinate the provision of safe delivery kits and child drugs in their health clubs. Drugs will be provided to treat the major diseases that cause child deaths in the community. The choice of drugs provided will reflect national and IMCI policies, and will be approved by the Ministry of Health. After each training block, the village health workers will return to their village and discuss the training in health clubs. They will also visit with mothers of children who have symptoms of disease, and discuss the training with them. Nurse-trainers will visit each village health worker twice per month to monitor the quality of their work, provide advice, and reinforce the training. An audit system will be maintained to keep track of the quality of the village health worker’s activities along with measuring the incidence of disease and recommended treatments in each community. The nurse trainers will also carry out monthly mobile clinic visits to villages. The services provided during these visits will include antenatal care, child health checks, and treatment for acute disease. During antenatal care visits, women will be assessed for risk, and those women that are considered to be at high risk will be highly encouraged to deliver in a hospital, and to travel to the hospital prior to delivery. These women will receive free primary and secondary services at the hospital. The trial will run for a period of two and a half years after the interventions have been fully implemented in the intervention clusters. All interventions will be given to the control clusters at the end of the trial if the research shows that the interventions are effective. The trial site is the region of Tombali and Quinara in the south of Guinea Bissau. The definition of clusters for the trial is complicated due to the lack of any accurate maps, village boundaries or recent population surveys. The last census was conducted in 1991, and since that time there has been substantial migration due to civil war and socioeconomic factors. The regional departments of health maintain lists of tabancas where they sometimes provide outreach services, but they do not have information on the population of these tabancas, nor maps showing accurate locations. One hundred and forty six clusters are to be enrolled in the trial. Clusters have been identified and enumerated during a baseline survey. Existing lists and maps of Tabancas used by the regional departments of health has been compiled to identify Tabancas. With the help of experts at the department of health, and visits to the tabancas prior to conducting the survey, all Tabancas that were believed to have at least 300 population were considered to be eligible. If two Tabancas with eligible size were less than 4 km apart we chose the Tabanca that has the smallest number of houses but possibly at least 40 houses. Major urban areas with more than 2000 population were excluded from the list. Fieldworkers travelled to the region and surveyed the populations in the listed Tabancas, or in unlisted Tabancas (if they found one not on the list once in the region). While the interventions will be made available to all households within a Tabanca, the analysis of the trial will be based on outcomes for the closest households to the centre of the Tabanca. The baseline survey aimed to enumerate an approximate population of 350 for each cluster. With an assumed average of 8 members per dwelling, the baseline survey aimed to track the nearest 40–50 dwellings to the health club meeting place. After defining a central point in the Tabanca, which was either the USC if it existed or any other central meeting place if not, field workers stayed in the village to map up to 52 but if achievable not less than 40 households. They interviewed all available eligible women within the households (see below). If there were fewer than 40 dwellings in the Tabanca the fieldworkers were asked to pick the nearest Tabanca within walking distance from the first Tabanca and to map additional households until they had reached a total of 40 but not more than 52 households. If they still did not reach 40 households, a third Tabanca was chosen nearby and this procedure continued. If there was no other eligible Tabanca within walking distance the fieldworkers were told to stop after mapping all available households. After surveying an adequate number of households to form a cluster (population approximately 350), field workers travelled to the next nearest eligible Tabanca to survey. This process resulted in 241 Tabancas forming 146 clusters being eligible for inclusion in the trial. The analysis of the trial will be based on outcomes for eligible women, and children living with them, and on the outcomes from new pregnancies arising during the trial. A woman is eligible for inclusion in the analysis of the trial if: • She lives in one of the 146 selected clusters. • She is a resident in one of the households mapped and is interviewed during the baseline survey. • She gives consent to participate in the trial. • She reports that she is between 12 and 49 years of age or/and she reports that she is the primary caregiver for a child under five years of age at the time of the baseline survey. • The tabanca she lives in gives consent to participate in the trial. Children are eligible for inclusion in the analysis if: • They are five years of age or younger at the time of randomisation and they reside permanently with an eligible women at the time of randomisation, and the child’s name was recorded during the baseline survey, or • They are born (live birth) to an eligible woman after baseline survey and are alive at the time of randomisation, or • They are born (live birth) to an eligible woman after randomisation. For the primary analysis, the survival of these children will be considered from 1 April 2008, by which time all main components of the intervention will have been implemented. This delay is necessary in order to give the implementation team time to register households for the intervention and to train personnel, and will also ensure that the interventions have sufficient time to have an impact on survival. According to this schedule, the interventions will start being implemented in control clusters at the end of September 2010, just after the end of the rainy season in the region (if the research shows that the interventions are effective). For the primary analysis, the survival outcomes of eligible children up to 30 Sept 2010 will be included in the analysis. Women and their children will be excluded from the analysis if they do not satisfy all inclusion criteria. Six mobile survey teams were trained to enumerate women in the mapped eligible dwellings in each cluster during the baseline survey. They recorded all eligible women living in these dwellings, and recorded their full birth history and child survival records. In the case of women who are normally resident in the tabanca but not present at the time of the baseline survey, the survey team recorded their name and house location but did not list their children nor record their birth history. There will be some women who are normally resident in intervention clusters who were not enumerated during the baseline survey. Since it would be unethical to exclude such women from access to health clubs and the services provided to other members of the village, they will be permitted to receive the interventions. Once randomisation is completed, the intervention team will enumerate all households in the intervention tabancas that had not previously been enumerated, and then invite the residents to take part in the intervention services. Only people who live in intervention Tabancas will be permitted to receive interventions, and women will be issued with a “mother’s card” to ensure that only women and their children who are eligible for the interventions will receive the interventions. Once this second enumeration is complete, no additional people will be permitted to take part in the interventions. To ensure that bias is not introduced, the women, and their pregnancies and children, in these households that were enumerated after randomisation will not be included in the primary analysis of the trial. The population may be affected by temporary and permanent migration. Primary analysis of the outcomes of the trial will be carried out according to the intention-to-treat principle, i.e. women will be analysed according to Tabanca of residence at the time of the baseline survey. Some people in the region may practise child-swapping: when a child stops breast-feeding, usually at 3–4 years of age, he or she moves to live permanently with the mother’s eldest brother and is looked after by one of the brother’s wives as foster mothers. Primary analysis of the trial will include any child permanently residing with an eligible woman at the time of the baseline survey, regardless of whether the child is the offspring of the eligible woman or not. In order to prevent possible bias, children who become fostered by an eligible women after randomisation will not be part of the trial. Women may temporarily move, or they may permanently move from a tabanca, in which case we may have difficulty tracking their child survival outcomes. We will try to learn outcomes from neighbours and friends but we will not attempt to personally interview women who have moved outside their enumeration cluster. When an eligible woman cannot be found, we will include the information from relatives, neighbours and other sources in the cluster about her pregnancy and survival outcomes, and her children’s survival outcomes, for the primary analysis. If a fieldworker did not meet with a woman in a prior monitoring, but they do meet with her on a subsequent visit, the field worker will ask the woman to verify the previously reported data that was received from relatives, neighbours or other sources. This trial will employ three levels of consent: • At the state level, approval of the protocol will be sought from the Ministry of Health in Guinea-Bissau. • At the Tabanca level, field supervisors will arrange meetings with local leaders to seek consent. In all tabancas in the trial, consent will be obtained from the leaders in the following manner. The local leaders will be informed about the protocol for the trial in their local language. They will then be asked to inform and seek consent, as appropriate for their ethnic group and specific traditions, from the members of their community. Consent will be given in oral form during this community meeting with written documentation of the discussions for verification. This process of obtaining consent from the ‘guardians’ of the clusters is common in trials in which the intervention is delivered at the level of a cluster and it is not possible to ask for prior consent for randomisation from individuals within the cluster [11]. • Given this process, women who are asked to take part in the quarterly enumeration and who consent to complete this enumeration are considered to have given their implied consent to participate in the trial. Women not living in the tabancas at that point, or who are not present, who subsequently fill the eligibility criteria will be informed about the study. The sample size has been calculated to give 80% statistical power to show a 30% reduction in the proportion of death before the age of five in the intervention clusters compared to the control clusters. We assume that the average size of the population in eligible households for each cluster will be 350 people defined by the algorithm described earlier. We also assume an average annual number of live births of twelve per cluster (the crude birth rate is estimated to be 35 per 1000 population in the region according to discussion with regional health authorities). Thus, we expect to observe approximately 48 children under five per cluster at the time of randomisation (taking into account the estimated probability of death at birth and consequent time intervals below 5 years of age based on a baseline survey conducted in Guinea-Bissau in preparation for this trial) and a further 30 births during the following two and a half years. We used an intraclass correlation coefficient (ICC) of 0.02 to allow for clustering. This was estimated from the DHS survey in India, using ten years data from Maharashtra state. There are no similar DHS data for Guinea-Bissau. In a Nepalese trial examining neonatal mortality, the ICC was estimated to be 0.00644, but since there were generally only one or two births per woman, and the clusters in that trial were much larger, we believe the ICC here will be higher. Under the above assumptions, the expected proportion of children who will die before the age of five during a two and a half year period in the control arm is 9%. Assuming 10% loss to follow up in both control and intervention arms, 130 clusters in total will be needed to show a 30% reduction in the proportion of deaths before the age of five in the intervention clusters compared to the control clusters with 80% statistical power and with 2-sided significance level of 5%. We are planning to map and randomise all the 146 clusters, defined as eligible during the baseline survey, to allow for possible dilution of the impact of the intervention that might be caused by migration. The main analysis will assess the effect of the interventions on the primary and secondary outcomes. Primary analysis of the outcome(s) will follow the intention to treat principle (i.e. the participants will remain in the group they were randomised to and not analysed according to the interventions actually received), and will account for clustering and stratification. For the primary outcome (proportion of children dying before the age of 5 years), the relative risk with a 95% confidence interval will be reported, taking appropriate account of clustering. A generalised linear model (with adjustment for stratification factors) using robust standard error estimates will be used to carry out this analysis. For secondary binary outcomes, including morbidity, maternal mortality and morbidity, institutional deliveries, home birthing practices, relative risks will be estimated in an analogous fashion. For continuous outcomes t-tests with robust SEs will be carried out. ANCOVA (with robust standard errors) will be used where baseline adjustment is required. Bootstrapping will be used for non-normal outcomes and Cox models for survival analysis. A full analysis plan will be given in the operations manual. The trial protocol has received ethical approval from the Ministry of Health, Department of Hygiene and Epidemiology Centre for Coordination of the Research in Guinea-Bissau (reference number: 021/2007) as well as from the ethics committee of the London School of Hygiene and Tropical Medicine (reference number: 5173).
N/A
