|
nif:isString
|
-
This study was conducted in the health district of Houndé, located 250 km west of Ouagadougou, Burkina Faso’s capital city. The location was chosen as we had previously carried out two randomized controlled trials in this district [32, 33] and our collaboration with the health services and population was effective. Houndé has a population of 294,865 inhabitants, one general district hospital, and 27 government primary health centers. At the time of the present study, each primary health center had at least three primary healthcare providers. Each primary health center offers a minimum package of preventive, promotional, and curative services for women and children. Preventive services include immunization (free for children under 5 years of age), deworming, and vitamin A supplementation. Promotional services include antenatal care, the Child Growth Monitoring and Promotion Program (GMP), and group communication on behavior change organized at the health center level. Such group communication may covered appropriate breastfeeding and complementary feeding practices, feeding of sick children, promotion of GMP session attendance, immunization promotion, insecticide-treated mosquito net use, common disease management (diarrhea, malaria, acute respiratory infection), antenatal care attendance, folic acid and iron supplementation during pregnancy, intermittent preventive treatment of malaria in pregnancy, pregnant women’s dietary needs, family planning, and hygiene and sanitation. Curative services include sick children consultations following the Integrated Management of Child Illness strategy. These services are continuous with a referral system to the general hospital. The average catchment area for primary health facilities had an average radius of 8.8 km in 2014, with 63% of the population living within 5 km of a facility [34]. In 2014, the average number of contacts with health centers was 1.8 per year for children under 5 years old in the district [34]. The health district has one pediatrician trained in nutrition responsible for monitoring all district nutrition activities. Treatment of acute malnutrition includes community-level passive case detection using mid-upper arm circumference measurements, referral to health centers for confirmation, and outpatient treatment of uncomplicated cases of severe acute malnourished children using ready-to-use therapeutic foods. Cases with medical complications or lack of appetite are hospitalized at the district hospital. Houndé had no specific services for moderate acute malnutrition cases at the time of this study (2008). Less than 6% of households in the region where the district is located are food insecure, against 19% nationally [35]. A situational analysis was conducted in the district in 2008, before implementation of the present intervention. This situational analysis comprised an in-depth investigation of all 27 primary health facilities to assess the functioning of maternal and child health services (observation of prenatal care, GMP sessions, and sick children consultations). After this detailed analysis, 12 health centers were selected as sites for the trial. The number of sites was limited by budget and logistical constraints. Selection was based on the availability of human resources able to be involved in the intervention delivery, and the availability of the minimum package of maternal and child services recommended nationally for first-line health facilities. A survey in a representative sample of 3968 children was conducted in the catchment areas of the 12 selected health centers to collect data on population characteristics, including the prevalence of stunting and wasting, feeding practices, and morbidity rates and their determinants in children under 5 years old. From this pre-study we observed that the average household size was 8±5 (mean±SD), with 2±1 (mean±SD) children younger than 2 years. Most women (86.6%) never attended school. The principle economic activity in the district was agriculture (96%). Nearly 32% of households used an unprotected well or pond as a source of water for consumption. Most households (67.8%) had no toilet or latrine and people often defecated in the open. In terms of breastfeeding and complementary feeding practices, 51.4% of children aged <6 months did not receive colostrum at birth, and the average duration of breastfeeding was 15.9±5.1 months. Complementary food was introduced before 6 months in 39% of children, with the principal complementary food being simple cereal porridge (85.2%). The mean dietary diversity score based on nine food groups was 2.0±1.2 for children aged 6–59 months. Furthermore, we found that 67.6% of pregnant women attending antenatal consultations did not receive prenatal dietary intake advice. In addition, 75.8% of caregivers attending healthy child consultations did not receive advice on complementary feeding. The pilot study also found that 16.4% of children under 5 years of age were wasted, 38.7% were stunted, and the overall morbidity rate was 38.7% (S1 Table). The average birth weight was 2,926±502 g, and the low birth weight prevalence was 14.1%.
We conducted a cluster randomized trial to assess the effect of preventive counseling during pregnancy and childhood on childbirth outcomes, child nutrition status, and morbidity. Healthcare workers were trained to offer counseling during their usual contacts with women. In the Burkina Faso health system, there are a limited number of healthcare providers in health facilities who interact with users daily. In addition, women attending the same health center usually live in the same community, and may communicate with each other and exchange experiences from that health center. Therefore, a cluster randomized design was the most appropriate design for the present study. A cluster was defined as the catchment area of a primary health center. All primary health centers that had functioned for at least 1 year at the time of the study were eligible (27 health centers in total). From these 27 health centers, 12 were selected to participate in this study. Selection was based on the availability of human resources in the health centers (at least three primary healthcare providers, as recommended by national policy), the actual existence and regularity of delivery of the minimum package of activities recommended at a national level for first-level health facilities (prenatal visits, healthy child consultations, immunization sessions, sick children consultations), and health center performance based on the number of children under 24 months who had attended a healthy child consultation the previous month and the rate of measles immunization in 2007 (S2 Table). As this research project was based on healthcare provider participation, the last criterion was healthcare providers’ willingness to be involved in a formative research project. This was assessed during group discussion with healthcare providers during an initial visit to each health center to present the research project. The 12 clusters were pair-matched using three criteria: i) distance to the district hospital; ii) mean household socioeconomic scores; and iii) child morbidity rate and prevalence of wasting and stunting. Data for the pair-matching was obtained from the 2008 pilot survey (S2 Table). In each pair, one health center was randomly allocated to the intervention arm, and the other to the control arm. Randomization was conducted publicly. For each pair of health centers, two identical pieces of paper were numbered corresponding to each health center and put into a basket. A volunteer not involved in the study was asked to choose a paper for the intervention center. After the first choice for the intervention center, the second center in that pair was systematically allocated to the control arm. No allocation concealment and blinding were possible. All participating centers, staff, and mothers were aware of the study arm to which they were assigned. Consent was obtained from each health center management committee, and a committee representative took part in the randomization session. No changes occurred in the trial conduct or outcomes after the study began.
The study population comprised pregnant women and their offspring living in the catchment areas of the 12 selected health centers for at least 6 months, and not planning to leave in the next 2 years. The primary outcome used to compare the two groups was the incidence of wasted children. As no data were available on the incidence of acute malnutrition in the study area before the intervention, the prevalence of global acute malnutrition (16.4%) found in children under 5 years old in the 2008 pilot study was used as a proxy. Sample size calculation was based on the hypothesis that the intervention would contribute to a reduction of 5% in the prevalence of wasting in children under 5 years old in the area. The coefficient of variation between clusters within the matched pairs was conservatively estimated at 0.25.We used the formula proposed by Hayes et al. for a pair-matched cluster design to calculate the required sample size [36]. Taking into account an estimated 20% of incomplete or missing observations, this calculation resulted in a sample size of 180 children per cluster or 1,080 subjects per study arm, using 80% power and 95% confidence interval (CI). To evaluate the intervention effect, a cohort of pregnant women in their third trimester was prospectively recruited from each cluster. Pregnant women were eligible for inclusion if they had no intention of leaving the study area for the next 2 years and provided informed written consent. Eligible pregnant women were identified through antenatal consultations (women attending their third antenatal visit), and were included until the desired sample size was reached in each health center. After birth, singleton babies without major birth defects were included in the study and followed quarterly until 18 months of age by external nurses not involved in the intervention delivery. Recruitment of study participants occurred from August 6, 2009 through December 29, 2011, and participants were followed until June 27, 2012.
In the control centers, routine preventive, promotional, and curative services were provided to pregnant and lactating women, and children aged <5 years as per national policy. Monthly growth monitoring and promotion (GMP) of all children below 24 months of age is organized by auxiliary nurses. During these session children are weighed and, if applicable, vaccinated (up to 9 months). The nutrition counseling intervention was implemented in the intervention centers within the usual care environment. The intervention aimed to: i) improve communication between care providers and women at any contact for prenatal visits and children’s services; and ii) enhance the nutrition component of the existing maternal and child national program, which includes prenatal care, immunization, and healthy and sick child consultations. Before the study, all healthcare providers in the intervention arm were trained in communication and nutrition at the district hospital in April 2009. The communication training focused on skills for educating caregivers, and how to listen to and build confidence with caregivers to introduce behavior change. Training was based on the patient-centered model developed by Stewart et al. [37], existing literature [38–41], and the World Health Organization (WHO)/United Nations Children’s Emergency Fund counseling course [42]. This training lasted 1 week and was led by a pediatrician with support from an expert on patient-centered communication from the Health Care Department, PXL University College, Belgium. Training covered three areas of knowledge: activation of medical knowledge, implementation of a framework for family anamnesis based on Gordon’s 11 functional health patterns, and basic communication models [43, 44]. The training used presentation materials, exchange of experiences, and role play. Finally, detailed family history sheets were developed, tested, and implemented by participants. The goal was to produce a tool that allowed extensive assessment of a family’s social structure, living conditions, and domestic habits. These sheets included medical and family background questions for specific pediatric problems related to malnutrition. Providers were also trained by the study team on maternal and child nutrition. This aimed to improve health providers’ skills in: (i) providing appropriate feeding counseling; ii) assessing child nutritional status and feeding problems; and (iii) making recommendations. This training was, focused on pregnant women’s diet, breastfeeding, complementary feeding practices, and further counseling skills based on the WHO Manual on Counseling the Mother and the Integrated Management of Child Illness strategy training manual [42, 45]. Particular attention was directed to communication, including imparting communication skills, detecting feeding problems, and negotiating with the caregiver on possible solutions that could be personalized using locally-adapted feeding recommendations. Two refresher training sessions were organized during the study period. Individual nutrition counseling was provided to all women attending the intervention centers during pregnancy and the first 18 months of their child’s life. Counseling contacts were scheduled to begin during pregnancy and continue until the child reached 18 months of age. At each contact, health providers used a patient-centered approach to explore the caregiver’s and child’s situation and current feeding practices, and identify their specific needs in terms of nutrition, health advice, and curative care. They then individually counseled caregivers on theirs needs or their child’s age-specific needs, taking into account the household socioeconomic situation and problems identified, allowing the caregiver to choose recommendations that were feasible and acceptable from various options. During pregnancy, counseling included the reduction of heavy work, advice on the need to rest, use of health services, consumption of iron and folic acid supplementation, and use of malaria prevention. Healthcare providers discussed dietary behavior and explored inadequate practices such as dietary prohibitions and restrictions related to some beliefs with pregnant women individually, using a patient-centered approach. They then oriented their advice to addressing the challenges and problems identified. General counseling themes included increased portion size and meal frequency, splitting meals, and introducing snacks. A second point of attention was dietary diversification, focusing on the consumption of iron-rich foods, protein-rich animal-source foods, or legumes such as beans and lentils. These messages were adapted to caregivers’ financial situations. Counseling on maternal dietary intake was maintained until the last postnatal consultation 6 weeks after delivery. During the prenatal period, healthcare providers provided counseling on appropriate breastfeeding practices, such as early initiation of breastfeeding, feeding colostrum, exclusive breastfeeding up to the age of 6 months, and timely introduction of complementary foods. After delivery, a patient-centered approach was used to identify specific individual needs and problems surrounding child feeding. Thereafter, tailored counseling was provided to promote appropriate feeding practices and offer solutions to any nutrition-related problems. Counseling on breastfeeding practices continued during early infancy. Additional focus areas were the timing of the introduction of complementary foods, the nutritional value of foods available at home, and the quality and quantity of complementary foods. After 6 months of age the nutrition counseling aimed to improve the quality and quantity of complementary foods. Specific counseling included how to blend food products available at home to improve a child’s dietary intake. Health providers asked caregivers to list all ingredients available at home and explained the benefits of each ingredient for the child. Based on the food available in the child’s household, health providers showed caregivers various recipes based on typical local meals and modified to increase their energy and nutrient density. For example, for porridge, a common complementary food usually made from cereal flour and water, caregivers were taught to add cooking oil or groundnut flour, soybean flour, or dried fish. Other topics included the improvement of food energy-density using household processes such as grain malting, improvement in food taste by fermenting cereals or adding sugar to porridge and so forth. Meat and egg consumption was promoted by showing caregivers the benefits of animal products for the child’s growth. During child curative consultations, healthcare providers explored the child’s ongoing feeding practices and tailored advice for the treatment of any illness. Finally, caregivers were counseled on recognizing symptoms of common childhood diseases (fever, diarrhea, acute respiratory infection, and so on), how to address these adequately, and how to recognize dangerous signs. The intervention was integrated into the health centers’ routine activities, and the same actors working for the existing health program were responsible for implementation. No additional topping up was paid to these healthcare providers for intervention delivery that was monitored by the district health team during routine quarterly supervision. The investigators’ role was restricted to the measurement of outcomes for assessment of intervention effects.
Data were collected by trained field workers not involved in the intervention delivery, and who were not blinded to the intervention. Health center appointments were made each month with participating caregivers. At the scheduled follow-up visits (at 3, 6, 9, 12, 15, or 18 months), caregivers were invited to attend the health center for data collection. The day before the data collection visit, a messenger went to the participating women’s homes to remind them of the appointment. Home visits were planned in the study protocol to encourage non-regularly attending mothers to bring their children to the health center, and to measure the actual care and feeding practices of children who had lost weight. However, for budgetary reasons, these visits were not effective. Caregivers’ behavior and practices were assessed using interviews with structured questionnaires. All questionnaires were pretested before the study started with a sample of mothers not included in the analysis. Questionnaires were administered in the caregiver’s local language. The first questionnaire was administered during the first month following birth and included information on the child and caregiver socioeconomic characteristics, dietary practices during pregnancy, and child delivery characteristics (e.g., whether the child was thin and presence of malformation). We added questions to assess early breastfeeding practices (delay before feeding the child after delivery, whether the child was fed colostrum, and feeding experiences in the first 72 hours). Women were asked questions about the follow-up of their pregnancy, including the number of completed antenatal visits, and whether they received preventive treatment for malaria and anemia during pregnancy. They were further questioned on their prenatal dietary practices, with emphasis on pregnancy-related changes in diet. The theme of culturally forbidden foods during pregnancy was also addressed. Finally, the adequacy of the intervention was assessed by asking caregivers whether they received counseling related to diet in pregnancy, early breastfeeding, how to breastfeed, and exclusive breastfeeding in the health center during their last contact with a healthcare provider for prenatal care. At each follow-up visit, caregivers were questioned about their knowledge on infant feeding, as well as the counseling on breastfeeding and complementary feeding they had received. Feeding frequency was assessed by asking caregivers to recall the number of times they fed their children in the 24 hours before the interview in addition to breastfeeding. Semi-quantitative 24-hour dietary recalls were conducted to assess child feeding practices. The caregiver was asked about all the foods the child had consumed since the previous morning and on the morning of the visit. Each consumed meal was deconstructed into foods. Each food was classified in a food group regardless of the amount consumed, unless the food in question was used as a condiment. Seven food groups were included: cereals, roots and tubers; legumes and nuts; dairy products (milk, yogurt, and cheese); meat products (meat, poultry, and offal) and fish; eggs; fruits and vegetables rich in vitamin A; and other fruits and vegetables. The response options were “yes, consumed” (score 1) and “no, not consumed” (score 0). During follow-up visits, caregivers were asked to recall their child’s morbidity history for the last 2 weeks using an adapted version of the generic Demographic and Health Survey instrument [46]. This was pre-tested in samples of about 20 mothers not belonging to the study in each health center (intervention and control centers).Caregivers who reported that their child had been sick were asked whether the child had suffered from one of the three main symptoms (diarrhea, fever, cough/breathing difficulties) or from other symptoms. Caregivers were questioned about their exposure to counseling on breastfeeding and complementary feeding practices in the health center during their last contact with a healthcare provider to assess the intervention adequacy at each follow-up visit. In addition, caregivers’ knowledge about danger signs for which she should bring the child to the health center was verified. The seven key signs that caregivers had to spontaneously mention were: the child cannot drink or suck, his/her health deteriorates, she/he develops a fever, the child displays fast breathing, the child has difficulty for breathing, presence of blood in the stools, and difficulty drinking. Infants’ birth weight was obtained from birth records, measured by health center nurses using baby scales available at the health centers. Health center nurses received training on anthropometric measurements and the baby scales were checked weekly to ensure accuracy. Children’s anthropometric measurements were taken at each visit following standard procedures [47]. Recumbent length was measured using length boards (Short Productions, Olney, MD, USA) with 0.1 cm precision. Infants were weighed in light clothing using electronic scales with a precision of 100 g (SECA 872, Germany). All measurement instruments were calibrated before each measurement session. All measurements were performed twice. Standardization exercises for anthropometric measurements were conducted during the initial training and were repeated quarterly during the study. At the time of supportive supervision, the degree to which healthcare providers in the intervention clusters applied the training that they received was assessed by the responsible district nutrition physician through observation and interviews with healthcare providers.
Primary outcomes were the cumulative incidence of wasting, and changes in child weight-for-height z-score (WHZ). Secondary outcomes were the women’s prenatal dietary practices, the Infant and Young Child Feeding practices indicators [48] (Table 1), the child’s birth weight, changes in child height-for-age z-score (HAZ), the prevalence of stunting at endpoint, and cumulative incidence of diarrhea, fever, and acute respiratory infection. Data were entered in double by two data clerks using Epidata [49].
Table data removed from full text. Table identifier and caption: 10.1371/journal.pone.0177839.t001 Definitions of indicators related to Infant and Young Child Feeding practices. Household socioeconomic scores were calculated based on assets and main housing characteristics, using the first component of principal component analysis and presented in quintiles [50]. Exposure to prenatal dietary counseling and early breastfeeding, how to breastfeed and exclusive breastfeeding was analyzed as a binary variable (yes/no) using data from the first follow-up visit. GMP attendance during the previous month and the exposure to counselling on complementary feeding practices during GMP sessions were also analyzed as a binary variable (yes/no) using data from the all the follow-up visits. Mother–child pairs were classified as “regular” or “non-regular” depending on whether or not they attended at least four visits (the median number of visits observed over the entire follow-up). A variable summarizing the level of exposure to counseling on complementary feeding was calculated. This variable was the percentage of GMP sessions where the mother received counseling on complementary feeding. This variable was further categorized into three levels: none (0%), low (0–33%) or high (34–100%). Dietary intake during pregnancy was coded as “0” for “no improvement” if the woman reported no dietary changes in quantity and diversity, and “1” if the woman reported an increase in meal quantity or diversity, or both. IYCF indicators were calculated from a qualitative 24-hour recall data. A total dietary diversity score was calculated using the recalled list of food items consumed over the previous day. These were summed to create a child dietary diversity score, ranging from 0 to 7. Minimal dietary diversity was defined as the consumption of at least 4 food groups. Low birth weight was defined as birth weight below 2500 g. WHO 2006 growth standards were used to calculate WHZ and HAZ using the ZSCORE06 Stata command [51]. Children with a WHZ below −2 were considered wasted and those with a HAZ below −2 were considered stunted [47, 52]. Morbidity was recalled over the two weeks before the interview. Diarrhea was defined as three or more liquid or semi-liquid stools in a 24-h period, fever was determined by caregiver report, and acute respiratory infection was defined as a combination of three symptoms (fever, cough, and respiratory distress). Each of the seven danger signs that required a child to visit a health center received a score of “1” if mentioned spontaneously by caregivers and “0” if not. Points were summed to calculate a knowledge score (range 0–7). All analyses were by intention-to-treat. Comparisons between study arms on participants’ baseline characteristics were performed using mixed-effects logistic regression. For the analyses of one point binary outcomes, we used mixed-effects logistic regression model with cluster pair as a random intercept and intervention nested as a random slope unless stated otherwise. For these analyses, results are reported as odds ratios (OR) and 95% CI. Linear mixed-effects models with cluster pair as a random intercept and intervention nested as a random slope were used to assess the effect of the intervention on continuous outcomes and binary outcomes collected at multiple point. The estimates were expressed as difference of mean or difference of proportions, and were presented with a 95% confidence interval (CI). All mixed effects models were adjusted for women’s age, parity, education level, and household socioeconomic score. For child related outcomes, we also adjusted for child age and sex. The effect of the intervention on the children’s continuous anthropometric outcomes (weight, length, WHZ, and HAZ) was analyzed using a linear piecewise mixed-effects model that included cluster pair and child as random effects. Knots were placed at 2, 4, 6, 8, 10, 12, 14, and 16 months of the child’s age. The linear piecewise mixed-effects model was preferred over a linear mixed-effects model after comparing the fit of both models using a likelihood ratio test. The better fit of the latter model was shown in a visual inspection of the graphs, which demonstrated non-linear relationship between child’s age and HAZ, and child’s age and WHZ. We tested the intervention effect on HAZ and WHZ by comparing the linear piecewise mixed effects model with interaction terms between the intervention allocation and the linear age spline terms against a model without interaction terms using a likelihood ratio test [53]. In cases of statistical significance, this would provide evidence of the effect modification over child’s age and therefore the existence of impact of the intervention on the trends of these continuous outcomes. We further tested the intervention effect on the cumulative incidence of wasting, caregiver-reported diarrhea, acute respiratory infection, and fever. We calculated incidence rates for both study arms with CIs derived from a Poisson distribution adjusted for clustering. Incidence rate ratios (IRR) were calculated using generalized linear latent and mixed models (GLLAMM). For this purpose, we fitted Poisson regression models with cluster pair and child as random effects using the Stata GLLAMM. A robust estimation of standard errors was used to relax the assumption of a Poisson distribution for binary data. For the longitudinal outcomes, children contributed person time to the analyses from birth until 18 months. For the morbidity outcomes, observation time was calculated by multiplying the number of effective visits with the recall period of two weeks, which was the reference period for the morbidity history recall at each visit. A child was defined as lost to follow-up at a given visit when he was absent from that visit and all subsequent visits. To clarify the number of children lost to follow-up and eliminate possible selection bias, an exploratory analysis was performed to compare sociodemographic characteristics between children declared lost to follow-up and those that contributed data until the age of 18 months. Missing data were processed as they were for the main analyses. However, to assess the robustness of our findings, we conducted multiple imputation (n = 50) of missing data using chained equations under the missing at random assumption, for the point outcomes: mother’s exposure to prenatal dietary counseling, counseling on early breastfeeding, counseling on how to breastfeed, counseling on exclusive breastfeeding, diet improvement during pregnancy, early initiation of breastfeeding, fed colostrum, received something else in the first 72 hours, child birth weight, and the endpoint prevalence of stunting. A P-value of 0.05 was considered to indicate statistical significance, and all tests were two-sided. All statistical analyses were conducted using Stata 12.0 (Statacorp, College Station, TX).
Eligible study participants were informed about the study aims and methods. Written informed consent was obtained from all enrolled mothers. The protocol was examined and approved by the Ethics Committee of the Ministry of Health of Burkina Faso (Comité d’éthique pour la Recherche en Santé), deliberation N° 2008–005 on 5 February, 2008. The Institutional Review Board of Tropical Medicine of Antwerp, Belgium later approved the protocol in Antwerp on 16 April, 2010 (ref: 10074712). Before implementation of the study, we obtained the approval of the local health authorities and the community representative on the health centers’ management committees. The protocol was registered at ClinicalTrials.gov with the identifier NCT01977365. However, an omission by the principal investigator meant that the study was registered after the enrolment of participants had started. All ongoing and related trials for this intervention have been registered
|
![[This page as RDF]](http://data.gesis.org/somesci/static/rdf-icon.gif){kind=icon}