Accounting for recent trends in the prevalence of diarrhoea in the Democratic Republic of Congo (DRC): Results from consecutive cross-sectional surveys

Objectives: To analyse trends in diarrhoea prevalence by maternal education, access to clean water and improved sanitation, household wealth index; to identify the sources of variation and assess contribution of changes in socioeconomic characteristics in the Democratic Republic of Congo (DRC). Design: Consecutive cross-sectional surveys. Setting: DRC. Participants: The databases contain information on 9748 children from the 2001 Multiple Indicators Cluster Survey and 7987 children from the 2007 Demographic and Health Survey. Interventions: N/A. Primary and secondary outcome measures: Whether the child had diarrhoea 14 days preceding the survey. Results: The overall prevalence of diarrhoea decreased by 26 percent (from 22.1% in 2001 to 16.4% in 2007). Findings from the three complementary statistical methods are consistent and confirm a significant decrease in diarrhoea regardless of socioeconomic characteristics. Changes in behaviour and/or in public health policy seem to be the likely main source of the change. There were no significant changes in diarrhoea prevalence associated with variation of the population structure. It is worth mentioning that the decrease in diarrhoea prevalence is in contrast to the generalised poor living conditions of the population. Therefore, it is difficult to ascertain whether the decline in diarrhoea prevalence was due to real improvement in public-health policy or to data quality issues. Conclusions: The decline of diarrhoea prevalence in our study need to be further investigated by conducting district-based or provincial-based studies to validate findings from household surveys such as Demographic and Health Surveys and Multiple Indicators Cluster Survey taking into account the current context of the country: ongoing conflict, poor socioeconomic and poor health infrastructure. However, improvement in living conditions such as access to clean water and improved sanitation will contribute to accelerate the reduction of diarrhoea prevalence as well as reduction of child mortality. This study uses two successive nationally representative household surveys: the 2001 Multiple Indicators Cluster Survey (MICS) and the 2007 DHS. During the 2001 MICS data collection from 21 May to 28 August 2001, three provinces were entirely under the control of the government (Kinshasa, Bas-Congo and Bandundu), four were partially administrated by rebels (Equateur, Katanga, Kasai-Oriental and Kasai Occidental) and four were entirely controlled by rebels (Oriental, Nord Kivu, Sud Kivu and Maniema). Though the 2007 DHS was carried out after the 2006 elections (2 February–30 April 2007 for Kinshasa, and from 10 May–31 August 2007 for the remaining provinces), some villages and municipalities in the Eastern provinces of Nord-Kivu, Sud-Kivu and Oriental were under armed conflict. The two datasets have comparable information on household characteristics and child diarrhoea at the time of the survey. The sample design and questionnaire are described elsewhere.8 9 Consequently, the two surveys offer the opportunity of analysing change in diarrhoea prevalence in the DRC. In total, the 2001 MICS database includes information about 8600 households and 9748 under-five children, whereas the 2007 DHS database had information about 8886 households and 7987 children. For each child under the age of five, the survey respondent in the household was asked whether the child has had diarrhoea in the past 2 weeks prior to the surveys as indicated in the Box 1 and in French language. Therefore, diarrhoea is determined not by medical examination but it is self-reported by the mother or caretaker with symptoms of three or more loose or watery stools per day or blood in stool. Exposure to diarrhoea-causing agents is frequently related to the use of contaminated water and to unhygienic practices in food preparation and disposal of excreta. Poor sanitation, lack of access to clean water and inadequate personal hygiene are responsible for an estimated 90% of childhood diarrhoea.3–5 13 Exposure variables for this study include maternal education, access to clean water and sanitation and household wealth index. A large body of empirical work has shown association between these variables and the prevalence of diarrhoea among under-five children.3 12–18 We define clean water or drinking water as water of sufficiently high quality that can be consumed or used with low risk of immediate or long-term harm. It is drawn from an improved drinking water source protected from outside contamination, in particular from contamination with faecal matter including piped water (into residence or plot), public tap, tube well, protected dug wells and protected springs.19 20 An improved sanitation facility is defined as one that is likely to hygienically separate human excreta from human contact: public sewer, septic tank, pour-flush latrine, pit latrine with slab, ventilated improved pit and ecological sanitation.19 20 The MICS and DHS surveys collecting these variables use the same definition and categorisation.21 22 In this study, household wealth index is measured with an asset index and wealth quintile constructed using the statistical Procedure Principal Component Analysis developed by Filmer and Pritchett.23 The index measures economic status based on housing characteristics, household assets and possession of household consumer durables as well as access to clean water and improved sanitation. The 2001 MICS and 2007 DHS have collected these data. Using rank methods, households are classified by quintile of wealth. This study uses three complementary methods: trends analysis, decomposition and longitudinal multivariate models (fixed effect regression models). The Stata, ‘nptrend’ command performs a non-parametric test of trend for the ranks across ordered groups. The test is an extension of the Wilcoxon rank-sum test.24 The test provides Z statistics and p value showing whether the change is statistically significant or not. The decomposition approach divides the trends in child’s diarrhoea prevalence into change in population structure and change in health behaviour and/or public health over the study period.25 26 This method assumes that the historical change in child diarrhoea prevalence depends on: (1) trends in distribution of under-five children by access to clean water and improved sanitation facility, household wealth index and maternal education over time (composition effect); (2) actual change in diarrhoea prevalence due to change in health behaviour or improvement in public health (the basic effect) that is the regression intercept when x=0 (α); (3) variation of diarrhoea prevalence by exposure variables (β), and the residual effect of other variables not considered as e error term (µ). This change can be presented as follows: The decomposition analyses are performed at an aggregated/cluster level (the national level by maternal education and household living conditions). Finally, we use a fixed-effect regression model to explore the relationship between women education and modern contraceptive use within the country. The equation for the fixed effects model is displayed below: where To perform the fixed effect models, we constructed three independent panel datasets (maternal education, access to clean water and improved sanitation and household wealth index). Each dataset has multiple observations about each category of the independent variable considered as individual (number of surveys, two in our case). Therefore, the maternal education database contains six observations, while numbers of observations for the access to clean water and improved sanitation and the household wealth index database are, respectively, estimated at 8 and 10 observations; each database contains the following information proportion of under-five children, year of survey and diarrhoea prevalence.