Health & Place 29 (2014) 84–94
Contents lists available at ScienceDirect
Health & Place journal homepage: www.elsevier.com/locate/healthplace
Spatial variation of management of childhood diarrhea in Malawi Dominic Odwa Atari a,n, Paul Mkandawire b a b
Department of Geography, Nipissing University, North Bay, ON, Canada The Institute of Interdisciplinary Studies, Carleton University, Ottawa, ON, Canada
art ic l e i nf o
a b s t r a c t
Article history: Received 29 January 2014 Received in revised form 24 May 2014 Accepted 19 June 2014
This study reports the spatial variability in household management of diarrhea among under-fives in Malawi. Using data from 2010 Malawi Demographic and Health Survey, we examined oral rehydration and feeding practices of mothers and caregivers of 3105 children with an episode of diarrhea by mapping district effect residual in geo-additive probit model and analyzing residual spatial effects in a Bayesian approach. The findings suggest that although diarrhea is relatively less prevalent in the Northern Region, this region lags behind in terms of adoption of appropriate practices for home-based management of diarrhea in children compared to the Central Regions and Southern Regions. A cluster of five predominantly rural districts in the eastern part of the Southern Region showed remarkably high level of household care for childhood diarrhea relative to the rest of the country. The fixed effects show the importance of breastfeeding, paternal education, wealth index, and ethnicity on oral rehydration, while paternal education, marital status, and ethnicity show significant influence on feeding for children with a diarrhea episode. The paper discusses the apparent inverse relationship between regional prevalence of diarrhea episodes and care-seeking practices for childhood diarrhea in Malawi, and makes relevant recommendations for policy. & 2014 Elsevier Ltd. All rights reserved.
Keywords: Diarrhea Children Spatial patterns Malawi Policy
1. Introduction Diarrhea is the passage of three or more loose stool or liquid per day. Diarrhea ranks amongst five most important causes of under-five morbidity and mortality in Malawi. In developing countries, diarrheal diseases account for an estimated 18–21% of all deaths in children younger than five years. This is equivalent to 1.5 million deaths per year (Boschi-Pinto et al., 2008). Nearly four in five of these deaths (78%) occur in Africa and South-East Asia. These two regions are also already excessively burdened by other infant and childhood diseases, including HIV/AIDS and malaria (Boschi-Pinto et al., 2008; UNAIDS, 2009). Although diarrhea can be dangerous as it can cause severe dehydration, simple household practices such as increasing the amount and frequency of drinks and food given to the child can replace the lost body fluids and salts and effectively treat the condition. Because treatment for diarrhea is relatively straightforward and inexpensive, health service providers, policy-makers, and researchers concerned with Integrated Management of Childhood Illnesses (IMCI) in poor countries seek to know about the spatial distribution of such child care practices in order to have a better understanding of how they can be promoted. Geographical
n
Corresponding author. Tel.: þ 1 705 474 3450x4196; fax: þ 1 705 474 1947. E-mail address: [email protected] (D.O. Atari).
http://dx.doi.org/10.1016/j.healthplace.2014.06.005 1353-8292/& 2014 Elsevier Ltd. All rights reserved.
location is thus an important factor in identifying clusters of the children who may be at elevated risk of diarrhea-related morbidity and mortality because of lack of awareness or poor adherence to these simple life-saving practices on the part of mothers or caregivers (Carter et al., 2000). Although there is expansive literature on childhood mortality and morbidity in Malawi in general, very few published studies have specifically focused on diarrhea. Available evidence however suggests that diarrhea-related under-five mortality and morbidity in Malawi is spatially highly uneven. A study conducted by Kandala et al. (2006), for instance, pointed to district-level socioeconomic and demographic (e.g. food security and population density) as important determinants of diarrhea-related childhood mortality. This study reinforced the findings of an earlier study conducted by Kalipeni (1993) which suggested that the district in which a child is born sets the context for risk for child morbidity and mortality. Kazembe (2007) also found dramatic variations in risk factors of co-morbidity related to fever, diarrhea and pneumonia in under-fives across regions in Malawi. This study also reported that age, habitual place of residence, nutrition status, and bed net use as important determinants, and urged more research to determine the precise linkages between childhood health and geographic location. In another study, Stockman et al. (2007) pointed to variations in access to safe and potable water as an important determinant of spatial differences in prevalence of under-five diarrhea. The authors found that rural poor women
D.O. Atari, P. Mkandawire / Health & Place 29 (2014) 84–94
are less likely to be aware of and to use simple diarrheal disease prevention techniques (e.g. point-of-use water treatment) relative to their urban counterparts. This study adds to this literature by examining both the suitability and adequacy of therapeutic responses of mothers and caregivers of children with known signs of diarrhea in Malawi. We specifically focus on whether there are significant geographic variations in childhood diarrhea management practices at the household level and, if so, the kind of factors that could explain such variations.
2. Context Malawi is located in the south-eastern region of the African continent (Fig. 1) with an estimated population of 14 million. Malawi is one of the poorest countries in the world with a per
85
capita income of US$383 and approximately 70% of Malawians live on less than one US$/day (Malawi Government, 2010). The majority of the population (80%) lives in rural areas and subsists on agriculture as smallholder famers. High levels of poverty are also linked to growing population pressure. Malawi was divided into three administrative regions after the country gained political independence from Britain in 1964; Northern Region, Central Region and Southern Region. The Northern Region comprises of five districts, Central Region has 9 districts, while the Southern Region has a total of 13 districts. The Southern Region is most populous, accounting for 45% of national population compared to 42% in the Central Region, and 13% in the Northern Region. High population growth rates relative to the fixed amount of landmass has meant that the population density in the Southern Region is nearly 3 times higher (185) than the Northern Region (63). The Central Region has a population density of 154 (Malawi Government, 2008).
Fig. 1. Study area.
86
D.O. Atari, P. Mkandawire / Health & Place 29 (2014) 84–94
Malawi has one of the highest under-five mortality rates in the world. According to the 2010 Malawi Demographic and Health Survey report, infant mortality is currently estimated at 66 per 1000 live births, down from 81 per 1000 live births in 2004. Similarly, a recent report from the United Nations named Malawi as one of the top 30 most dangerous places for children under-five, with an overall mortality rate of 92 deaths per 1000 live births in 2010 (in comparison, Canada and most European countries have rates between 2 and 6) (UNICEF, 2012). This is down from 145 per 1000 live births in 2004 (Malawi Government, 2010). These improvements in child survival are in part the result of policy of Integrated Management of Childhood Illnesses (IMCI) which has been implemented by the government in collaboration with UNICEF and WHO since 1999. A key component of IMCI is community integrated management of childhood illnesses which involves improving the ability of mothers and caregivers at the household and community levels to promptly recognize, treat and refer suspected cases of malaria, diarrhea, measles, acute and respiratory infections in under-fives. Despite apparent improvements, the current rate of under-five mortality is still excessively high. The 2010 MDHS report states that gender relations, birth order, mother’s education, and place of residence are important determinants of child mortality and morbidity in Malawi (National Statistical Office (NSO), ICF Macro, 2011). Access to essential health care is also low. For instance, only 80% of children are fully vaccinated, with those rural areas more likely to be vaccinated at 82% than those in urban areas at 76% (Malawi Government, 2010). Nearly one in every 10 children under the age of five present symptoms of acute respiratory infection in Malawi, but only seven out of every 10 of these children receive appropriate treatment within 24 h of onset of the signs. Similarly, prompt treatment for presumed malaria among under-fives remains low. Malawi Demographic and Health Survey (MDHS, 2010) shows that only 61% of under-five children presenting fever where taken to a health facility for treatment in the two weeks prior to the survey. There are considerable regional and district variations in factors related to the risk and management of diarrheal conditions in Malawi. For instance, levels of educational attainment vary dramatically across the three administrative regions. Due to early exposure to Scottish missionary education, the Northern Region’s historic advantage in functional literacy over the other regions continues into the present day. For instance, nearly 90% of men and 80% of women can read and write in the Northern Region, compared to 80% among men and 60% among women in the Southern Region. Beyond population densities and literacy rates, the three regions in Malawi also differ on a number of other relevant dimensions. Despite having high level of literacy, the Northern Region is politically weak and significantly underdeveloped compared to other regions. A recent welfare monitoring report revealed significant regional and district level variations in access to social amenities, as measured by a number of variables including ease of access to public transportation or hospitals. The report indicated, for instance, that 22% of household in the Northern Region live within less than an hour of the nearest public transport port compared to 33% in the Southern Region (Malawi Government, 2012). In the eastern districts of the Sothern Region, such as Thyolo, access to public transportation is much higher, with over 50% of the population living within less than 1 h of a public transport port. Similarly, only 68% of households in the Northern Region live within a 15 min-walk to the nearest source of drinking water, compared to 77% in the Southern Region. In eastern districts of the Southern Region, such as Phalombe, more that 82% of the households are located within a 15 min-walk to the nearest source of water. In terms of household practices for prevention of diarrheal diseases, such as boiling drinking water,
there is a clear north–south gradient with only 15% of households in the Northern Region reporting boiling water before drinking, compared to 20% in the Central Region and 25% in the Southern Region. In some districts in the eastern part of the Southern Region, such as Thyolo, the proportion of households reporting boiling drinking water is high as 36%.
3. Data and methods The study applied Bayesian hierarchical model and geostatistical techniques with location (district), and nonlinear metrical (mother’s and child’s age) attributes and other information to gain a better understanding of the spatial variation in the management of diarrhea in Malawi. The study uses the 2010 MDHS, the most recent nationally representative dataset of demographic and health collected by Malawi Government in collaboration with MEASURE (National Statistical Office (NSO), ICF Macro, 2011). A more detailed explanation of the data collection and methodological procedures followed by the 2010 MDHS are explained elsewhere (National Statistical Office (NSO), ICF Macro, 2011), but we highlight that the 2010 MDHS data were collected using a twostage stratified sampling design to provide national, regional and district estimates of health and demographic indicators. In the first stage, 849 enumeration areas, or clusters, were selected as primary sampling units, stratified by urban/rural status using sampling probability proportional to the population of the selected cluster. In the second stage, a fixed number of households were randomly selected from each cluster. A total of 13,220 women of reproductive age range (15–49 years) were interviewed using an interviewer-administered questionnaire which included questions capturing information on birth histories. A sub-sample of 3105 records was drawn involving all cases where a mother or caregiver reported an episode of child under-five diarrhea in the two weeks prior to the survey for use for the present study. The 2010 MDHS asked respondents “Has your child had diarrhea in the last 2 weeks?”. If the mother answers yes, a follow-up question was asked as to whether she provided their sick child with “less than usual”, “the same as usual”, or “more than usual” amount of fluids (oral rehydration) and food (feeding). Theoretically relevant predictors including individual variables (child’s sex, birth order, age, and breastfeeding condition), family (mother’s age, parental education, number of children under five, marital status, wealth index, and household size), and community (place of residence, ethnicity, and district of residents) characteristics were examined. Birth order was categorized into ‘1st’, ‘2nd and 3rd’, ‘4th and 5th’, and ‘6 or higher’, and breastfeeding status was dichotomized into ‘Yes’ for mothers who were breastfeeding at the time of the study and ‘No’ for those who were not. Maternal and paternal educational attainment was categorized as ‘no education’, ‘primary education’, and ‘secondary or higher education’. The number of children under the age of five in the household was categorized as ‘1 child’, ‘2 children’, and ‘3 or more children’. Furthermore, marital status was dichotomized into two categories; ‘married or living together’ and ‘Single’ mothers (never married, widowed, divorced, and separated). The 2010 MDHS also has a wealth index which captures the long-term standard of living based on household’s ownership of consumer goods; dwelling characteristics; type of drinking water source; toilet facilities; and other relevant characteristics related to a household socioeconomic status (National Statistical Office (NSO), ICF Macro, 2011; Gwatkin et al., 2003). Wealth quintiles included ‘Poorest’, ‘Poor’, ‘Middle’, ‘Rich’ and ‘Richest’. Household size was categorized as ‘small’ (if there were fewer than 5 members), ‘medium’ (if there were 5–9 members), and ‘large’ (more than 10 members). Place of residence was dichotomized into
D.O. Atari, P. Mkandawire / Health & Place 29 (2014) 84–94
87
Table 1 Distribution and univariate ordinal regression analysis of factors in childhood diarrheal management in Malawi (MDHS, 2010). Individual characteristics
Sex of child Male Femaleb Birth order 1st 2nd and 3rd 4th and 5th 6þ b Child’s age in monthsa o 6 months 6–11 12–23 24–35 36 þ Monthsb Breastfeeding No Yesb Family characteristics Mother’s age in yearsa o21 Years 21–24 25–29 30–34 35–39 40 þYearsb Maternal education No education Primary education Sec. education or higherb Paternal education No education Primary education Sec. education or higherb Number of under fives 1 child 2 children 3 þ childrenb Marital Status Married/living together Single mothersb Wealth index Poorest Poor Middle Richer Richestb Household size Small ( o 5 members) Medium (5–9 members) High ( 49 members)b Community characteristics Place of residence Urban Ruralb Ethnicity Chewa Tumbuka Lomwe Tonga Yao Sena Nkonde Ngoni Amanganja/Anyanja Others
Count (%)
Rehydration
Feeding
Estimate
95% CI
Estimate
95% CI
1575 (50.7) 1530 (49.3)
0.03
0.05
0.11
0.01
0.10
0.08
657 (21.2) 1134 (36.5) 718 (23.1) 596 (19.2) 21.4 713.6 147 (4.7) 750 (24.2) 1104 (35.6) 582 (18.7) 522 (16.8)
0.05 0.04 0.01
0.08 0.07 0.14
0.18 0.16 0.11
0.12 0.13 0.01
0.02 0.01 0.14
0.25 0.25 0.13
0.01 0.08 0.02 0.01
0.20 0.21 0.10 0.14
0.22 0.05 0.13 0.12
0.65 0.20 0.23 0.02
0.90 0.33 0.35 0.15
0.40 0.07 0.10 0.13
0.01
0.11
0.08
0.16
0.06
0.25
0.03 0.03 0.05 0.01 0.03
0.22 0.15 0.13 0.18 0.23
0.16 0.20 0.22 0.19 0.17
0.10 0.06 0.02 0.06 0.15
0.30 0.13 0.20 0.26 0.36
0.10 0.24 0.17 0.13 0.07
484 (15.6) 2225 (71.7) 396 (12.8)
0.16 0.03
0.31 0.16
0.01 0.09
0.10 0.05
0.26 0.18
0.07 0.08
406 (13.1) 1924 (62) 775 (25)
0.16 0.16
0.29 0.26
0.02 0.07
0.10 0.12
0.24 0.23
0.05 0.02
1144 (36.8) 1492 (48.1) 469 (15.1)
0.08 0.01
0.05 0.11
0.20 0.13
0.05 0.05
0.08 0.18
0.18 0.08
2678 (86.2) 427 (13.8)
0.06
0.06
0.18
0.05
0.08
0.17
0.21 0.22 0.07 0.07
0.35 0.36 0.21 0.22
0.06 0.08 0.07 0.08
0.03 0.13 0.03 0.03
0.18 0.28 0.12 0.18
0.12 0.03 0.18 0.13
1018 (32.8) 1883 (60.6) 204 (6.6)
0.06 0.01
0.11 0.16
0.23 0.17
0.06 0.06
0.12 0.24
0.24 0.12
290 (9.3) 2815 (90.7)
0.15
0.01
0.28
0.11
0.03
0.26
1080 (34.8) 319 (10.3) 490 (15.8) 49 (1.6) 327 (10.5) 151 (4.9) 46 (1.5) 400 (12.9) 145 (4.7) 98 (3.2)
0.36 0.24 0.65 0.17 0.33 0.41 0.18 0.68 0.45
0.11 0.03 0.39 0.23 0.06 0.11 0.62 0.42 0.15
0.60 0.51 0.90 0.58 0.60 0.71 0.25 0.95 0.75
0.18 0.30 0.38 0.22 0.38 0.56 0.63 0.32 0.25
0.09 0.01 0.11 0.21 0.09 0.24 1.17 0.04 0.08
0.45 0.59 0.66 0.65 0.67 0.87 0.08 0.60 0.58
786 (25.3) 2319 (74.7) 27.3 7 6.8 487 (15.7) 782 (25.2) 816 (26.3) 500 (16.1) 315 (10.1) 205 (6.6)
767 (24.7) 695 (22.4) 723 (23.3) 531 (17.1) 389 (12.5)
Rehydration and feeding are ordered ordinal variables (1 ¼less than usual, 2 ¼same as usual, 3¼ more than usual). a b
Mean 7 Standard deviation. Reference category.
‘Rural’ and ‘Urban’. Malawi is an ethnically diverse country comprising of 10 major tribes; Chewa, Tumbuka, Lomwe, Tonga, Yao, Sena, Nkonde, Ngoni, Mang’anja, and Nyanja. This study
therefore also controlled for ethnicity. The distributions of variables analyzed in the management of childhood diarrhea in Malawi are outlined in Table 1.
88
D.O. Atari, P. Mkandawire / Health & Place 29 (2014) 84–94
Fig. 2. Spatial distribution of observed diarrheal prevalence by regions (left) and districts (rights) in Malawi.
3.1. Statistical analysis The dependent variable is defined as yi ¼1 if child i had received oral rehydration or feeding that was less than usual, 2 same as usual, or 3 more than usual. The commonly adopted model for the analysis of this type of data is the probit or logistic model, and the standard measure of effects is the estimated mean or odds ratio (OR) (Mbonye, 2004; Woldemicael, 2001). However, because of the spatial nature of our data and the presence of nonlinear effects for some covariates, the assumption of a linear relationship between predictors and the dependent variable may not hold leading to biased parameter estimates. Hence, a flexible
Geo-additive logistic model was used (on the probability of a child receiving sufficient rehydration or feeding) to determine the socioeconomic and demographic variables associated with management of diarrhea while simultaneously controlling for spatial dependency in the data and possible nonlinear effects of the independent variables. Both oral rehydration and feeding were examined as ordinal dependent variables. Due to the ordered ordinal nature of the dependent variables (oral rehydration and feeding), the following cumulative probit model was used:
yi
(r)
( f1(xi1)
+ fp(xip) + f spat (si
i
it
ð1Þ
D.O. Atari, P. Mkandawire / Health & Place 29 (2014) 84–94
Table 2 Regions and districts of Malawi. Region and districts Region Northern
Central
Southern
Count (%) District Chitipa Karonga Nkhatabay Rumphi Mzimba
N ¼ 3105 77 (2.5) 82 (2.6) 36 (1.2) 109 (3.5) 124 (4.0)
Total Kasungu Nkhota Kota Ntchisi Dowa Salima Lilongwe Mchinji Dedza Ntcheu
428 (13.8) 222 (7.1) 134 (4.3) 136 (4.4) 128 (4.1) 126 (4.1) 162 (5.2) 122 (3.9) 131 (4.2) 92 (3.0)
Total Mangochi Machinga Zomba Chiradzulu Blantyre Mwanza Thyolo Mulanje Phalombe Chikwawa Nsanje Balaka Neno
1253 (40.4) 83 (2.7) 128 (4.1) 129 (4.2) 89 (2.9) 120 (3.9) 90 (2.9) 96 (3.1) 119 (3.8) 195 (6.3) 70 (2.3) 115 (3.7) 94 (3.0) 96 (3.1)
Total
1424 (45.9)
where Φ is the standard normal cumulative distribution function, yi A {1, 2, 3} denotes the rehydration or feeding which child i received, and f1, …, fp are nonlinear smooth effects of covariates (e. g. child’s and mother’s age) while fspat is the effect of district si A {1, …, S} where child i lives. The standard measure of effect in the probit model is the mean. However, because we used the Bayesian approach that relies on prior assumptions to make posterior inference, we used the posterior mean instead. To account for possible departures from the assumed distribution, 95% confidence intervals (CIs) for the posterior odd ratios (ORs) and probability maps (the equivalent of CIs for the spatial effects) are calculated using robust standard errors estimated via Markov Chain Monte Carlo (MCMC) simulation techniques. The analysis was carried out using BayesX-version 2.1 (Belitz et al., 2012), a software for Bayesian inference based on MCMC simulation techniques. The posterior distribution is intractable, so MCMC algorithms are used to generate samples from this prior distribution, which allows estimation and inference for the parameters. For each of the rehydration and feeding models considered in this study, 12,000 iterations were carried out with a burn-in period of 2000. We also investigated sensitivity to choice of priors by varying the values of hyper-parameters a and b and the results are stable for a ¼b¼0.001. This is consistent with what other authors have experienced (e.g. Adebayo and Fahrmeir, 2005; Kazembe, 2009; Gayawan and Adebayo, 2013).
4. Results The results show that overall 45.6% of under-five children received less than usual amount of oral rehydration while 33.6% and 20.7% received the same or more than usual amount of fluids
89
during a diarrhea episode, respectively. In terms of feeding, 64.2% of under-fives were fed less than usual amount of food during a diarrhea episode compared to only 28.9% and 7.2% who were provided with the same or more than usual amount of feeding, respectively. Table 1 shows the distribution and univariate ordinal regression analysis of factors related to childhood management of diarrhea by caregivers in Malawi. The sample had approximately equivalent number of male (50.7%) and female children. More than 21% of the children were first order births while 19% were at least the 6th order births. The mean age of the children was 21.4713.6 months. The average age of mothers of children was 27.376.8 years. Nearly 37% of mothers had 1 child under the age of five, while 48% and 15% of mother had two, and three or more underfive children, respectively. Approximately 25% of children were still breastfeeding at the time of the survey. With regards to education, only 13% of mothers completed secondary school or higher. The univariate ordinal regression results (Table 1) indicate that mother’s age, paternal education, wealth index, place of residence, and ethnicity are significant determinants of oral rehydration at the household level, while birth order, child’s age, breastfeeding condition, paternal education, wealth index, and ethnicity show significant influence on feeding. Fig. 2 shows spatial variations in the prevalence of diarrhea in Malawi. The results indicate clear disparities. Regionally, the Northern Region shows the lowest prevalence, followed by the Southern Region and the Central Regions (Fig. 2a). At the district level, highest prevalence was found in the Central Region district of Kasungu and the Southern Region district of Phalombe (Fig. 2b). Table 2 indicates the distribution of the study subjects amount the regions and districts of Malawi. The results for oral rehydration, as presented in Fig. 3, suggest considerable spatial auto-correlation in the underlying posterior means. The left panel reveals a cluster of high-risk practices by mothers who provide less than usual amount of fluids to their children during a diarrhea episode in the Northern Region districts of Chitipa, Karonga, Rumphi, and Mzimba compared to mothers in the southern districts of Zomba, Phalombe, Mulanje, Chiradzulu, and Thyolo who were more likely to provide more than usual rehydration to their children in case of diarrhea. The results of the nonlinear effect of child’s age (Fig. 4a) on oral rehydration suggest that there was an increasing potential for mothers to provide more oral rehydration to their children when they were younger than 26 months after which it levels off. We show estimated posterior means together with 95% CIs. Although there is a general tendency for mothers to rehydrate their children with increasing age, the influence of mother’s age on child rehydration was also nonlinear (Fig. 4b). With regard to the fixed parameters, Table 3 shows that children who were not being breastfed were also less likely to be orally rehydrated more than usual compared to those who were breastfed during diarrhea episodes. Children whose fathers completed only primary education were less likely to receive more than usual amounts of oral rehydration compared to those whose fathers had university training. With respect to poverty, women in the poorest or poor household were reportedly less likely to provide higher amounts of oral rehydration to their kids with diarrhea compared to those in the richest households. Children of the Ngoni ethnic group were more likely to get more than usual rehydration treatment when they have diarrhea than children of other ethnic groups in Malawi. We did not find any statistically significant associations between the likelihood of providing more than usual amount of oral rehydration and child’s sex, birth order, maternal education, number of children under-five, marital status, household size, and place of residence.
90
D.O. Atari, P. Mkandawire / Health & Place 29 (2014) 84–94
Black – significant positive effect Grey – Not significant White – significant negative effect
Fig. 3. Residual spatial districts effects (left) and 95% posterior probability map (right) of childhood rehydration during diarrheal episodes in Malawi.
Similar to the spatial distribution of oral rehydration, Fig. 5 shows that children in the Northern Region districts of Chitipa, Karonga, Rumphi, and Mzimba and Central Region district of Nkhota-Kota were less likely to be fed more than usual amount of food compared to children in the Southern Region districts of Mulanje, Chiradzulu, Thyolo, and Nsanje who were more likely to receive more than usual amount of food when they have diarrheal episodes. Children in the other districts showed no significant differences in their tendency to be fed more than usual amount of food during diarrhea. Table 4 shows the fixed parameters with paternal education showing significant influence on children feeding. Married mothers or those living together with their partners were more likely to feed their sick child with more than usual amount of food. Children of the Tumbuka ethnic group were more likely to be fed more than usual amount of food when they have diarrhea compared to children of other ethnic groups in Malawi. Fig. 6a shows the nonlinear effect of child’s age on feeding. There is an increasing tendency for children under 30 months old to be fed more than usual amount of food when they have
diarrhea after which the age effect levels off. Overall, mother’s age indicate nonlinear effects on child’s diarrheal management (Fig. 6b).
5. Discussion and conclusions This study has shown significant district-specific geographical variation in the management of childhood diarrhea at the household level in Malawi. When compared, the north–south pattern of diarrhea, prevalence remains higher in the Northern Region relative to the Southern Region (Fig. 2a). The posterior mean estimates of the residual smooth spatial district effects are shown in the left panel of maps of Figs. 3–5. In addition, posterior probability maps (right panel of Figs. 3–5) indicate significance of the spatial effects (shades of white colored¼negatively significant, shades of black colored¼positively significant, and shades of grey colored ¼not significant). Note that the residual spatial effects are centered about zero, i.e. the average of all districts is zero.
D.O. Atari, P. Mkandawire / Health & Place 29 (2014) 84–94
91
Table 3 District posterior mean estimates of the fixed effect parameters for childhood rehydration during diarrheal cases in Malawi.
0.3 0.2
Variable 0.1 0 -0.1 -0.2 -0.3 -0.4
1
6
11
16
21
26
31
36
41
46
51
56
Child's age in months
0.6 0.4 0.2 0
Constant Breastfeeding No Yesa Paternal education No education Primary education Secondary education or highera Wealth index Poorest Poor Middle Richer Richesta Ethnicity Chewa Tumbuka Lomwe Tonga Yao Sena Nkonde Ngoni Amanganja/Anyanja Othersa
Mean
Std. error
95% CI
0.02
0.21
0.42
0.40
0.13
0.06
0.25
0.00
0.11 0.13
0.08 0.05
0.26 0.24
0.04 0.02
0.19 0.22 0.08 0.05
0.09 0.08 0.08 0.08
0.36 0.39 0.25 0.21
0.02 0.06 0.08 0.11
0.28 0.31 0.22 0.21 0.13 0.17 0.11 0.48 0.10
0.16 0.16 0.17 0.24 0.17 0.21 0.26 0.17 0.18
0.02 0.01 0.10 0.25 0.20 0.22 0.40 0.16 0.25
0.60 0.64 0.55 0.66 0.47 0.58 0.62 0.81 0.46
-0.2 a
Reference category.
-0.4 -0.6
15
20
25
30
35
40
45
Mother's age in years Fig. 4. Nonlinear effects of child’s and mother’s age on childhood rehydration.
Over and above the impact of the fixed effects, there appear to be widespread negative influences on oral rehydration and feeding of children with diarrhea in the Northern Region districts. Remarkably, despite high levels of functional literacy compared to national average, children with diarrheal episodes in the Northern Region received less than recommended amounts of oral rehydration and feeding relative to children in the Central Region and Southern Region districts. Poor management of diarrhea in the Northern Region districts however exists within an epidemiological context of widespread low prevalence of the disease compared to the other regions (Table 2) (National Statistical Office (NSO), ICF Macro, 2011). This means that episodes of childhood diarrhea in this region are relatively low relative to the other regions, yet when the disease occurs, caregivers may not have appropriate level of knowledge and skills to properly manage it at home. This may be due to a number of reasons. First, low prevalence of diarrhea in the Northern Region may be diverting policy attention towards those areas where the disease is most prevalent. Second, in a context where resources are limited, prioritizing high prevalence areas is understandable. Third, the region’s low population density, with most of its population dispersed over a rugged topography with poor feeder road network has meant that efficient delivery of health programs is always a challenge. For instance, districts such as Chitipa, Rumphi, and Nkhata-Bay are mountainous with large swaths of hard-to-reach populations, in some cases only reachable by boat. Finally, widespread poor health-seeking behaviour for diarrhea in the Northern Region may be linked to historical factors where, lacking political power, the districts in the region find it relatively difficult to attract
government resources and policy attention as compared to the districts from the other regions. Unlike the Northern Region, Central Region districts show no significant effect in childhood diarrhea management. This intermediate status may correspond to a range of political, socioeconomic and environmental factors that characterize the region. For instance, while relatively high population density and mobility, and rapid urbanization may foster the spread of diarrhea, being the region that houses the capital city has always meant that the Central Region has always enjoyed a level of access to government resources and policy that is unknown to the other regions (Posner, 2004). In addition to political clout, the Central Region’s relatively less challenging topography may make it attractive for organizations to provide health education and delivery health services even in rural remote areas as evident by the high number of NGO in most districts in the region. Thus, geographic factors in the region may interact with social and political factors to keep diarrhea in check, as opposed to the Northern Region. High level of population density, increased population mobility, and sprawled conditions in small towns and estates in the Southern Region are well-known risk factors for the spread of diarrhea (Malawi Government, 2012). At the same time, large swaths of densely populated areas with people living in close proximity in the agricultural frontiers and rural growth centers of the region may facilitate rapid diffusion of health messages and practices. In addition, high population density can also lead to dense social networks which may have the secondary effect of creating a more supportive environment for adoption and sustainment of positive household health-seeking behavior (Wakefield et al., 2010). Remarkable improvements in the care of diarrhea exhibited by a cluster of five Southern Region districts of Zomba, Mulanje, Chiradzulu, Thyolo, and Phalombe should also be understood within the context of the growing political attention that these districts have garnered over the past two decades and an increase in resource flows to those districts that has accompanied the attention. These five districts account for about 20% of Malawi’s
92
D.O. Atari, P. Mkandawire / Health & Place 29 (2014) 84–94
Black – significant positive effect Grey – Not significant White – significant negative effect
Fig. 5. Residual spatial districts effects (left) and 95% posterior probability map (right) of childhood feeding during diarrheal episodes in Malawi.
population and constitute the strongest political base for all the three state presidents that have ruled Malawi since multiparty democracy in 1994. Concerted government policy in these districts over the decades, coupled with increased service delivery by NGOs and other civil society organizations may explain the remarkably high level of positive health behavior in household management of diarrhea, despite a widespread low level of literacy. The fact that these five largely rural districts outperformed even Blantyre, the country’s commercial capital city, gives hope that the apparently onerous task of reducing diarrhea-related morbidity and mortality in Malawi is not an impossible project. The example demonstrated by the five districts suggests that, given the right conditions, communities and households can be empowered to successfully prevent and treat cases of diarrhea in their children and therefore stop unnecessary deaths from occurring. For instance, the government could potentially capitalize on already high rate of functional literacy in the Northern Region to intensify health literacy, thereby enhancing the ability of mothers in this region to care for their children during an episode of diarrhea. In addition, investment in transport and communication
infrastructure can reduce logistical challenges that currently undermine effective health service delivery in various parts of the country, especially the Northern Region. This can help to expand coverage and facilitate scale-up of health services to hard-to-reach communities, especially those in hard-to-reach areas of the Northern Region which may be ill-equipped to deal with diarrhea due to being unfamiliar with the disease or lack of prior exposure to relevant health education and skills. In terms of individual-level parameters, the results indicate that the effects of child’s and mother’s age on the amount of oral rehydration and feeding were predominantly nonlinear (Figs. 4–6). The analysis shows that older children are more likely to be provided with adequate amounts of oral rehydration and feeding compared to their younger counterparts. The general tendency for the frequency of child feeding to increase during an episode diarrhea with the mother’s age may in part explain a known trend of increasing childhood morbidity among young mothers who are often not only economically disadvantaged, but also less experienced in terms of child caring relative to older mothers Kandala and Madise, 2004)
D.O. Atari, P. Mkandawire / Health & Place 29 (2014) 84–94
Table 4 District posterior mean estimates of the fixed effect parameters for childhood feeding during diarrheal cases in Malawi. Variable
Mean
Std. error
95% CI
Constant Paternal education No education Primary education Secondary education or highern Marital status Married/living together Single mothersn Ethnicity Chewa Tumbuka Lomwe Tonga Yao Sena Nkonde Ngoni Amanganja/Anyanja Othersn
0.53
0.22
0.95
0.11
0.06 0.12
0.08 0.06
0.22 0.24
0.10 0.00
0.15
0.06
0.04
0.26
0.15 0.40 0.10 0.21 0.20 0.17 0.50 0.20 0.01
0.17 0.17 0.18 0.25 0.18 0.22 0.32 0.18 0.20
0.18 0.06 0.25 0.29 0.16 0.26 1.13 0.13 0.37
0.48 0.74 0.45 0.71 0.56 0.61 0.11 0.55 0.39
0.6 0.4 0.2 0 -0.2 -0.4 -0.6 -0.8 -1 -1.2
1
6
11
16
21
26
31
36
41
46
51
56
Child's age in months
93
to children (Millard, 1994). For instance, while examining child morbidity in Malawi and Zambia, Kandala and Madise (2004) found that childhood morbidity was lower among educated women and lower morbidity was also reported in households with large number of adult members. In addition, it is not surprising that breastfeeding emerged as an important factor for oral rehydration. IMCI policy guidelines stipulate that mothers of children with diarrhea should be encouraged to breastfeed their children more frequently as part of oral rehydration. One of the major reasons why exclusive breastfeeding is encouraged in developing countries relates to the need to reduce the risk of diarrhea in infants. This means that mothers who practice breastfeeding are more likely to know about its benefits in relation to diarrhea case management in children (Bhandari et al., 2003). There are also ethnic differences in both rehydration and feeding for diarrheal management in Malawi. For example, children from the Ngoni and Tumbuka ethnic groups were more likely to be sufficiently rehydrated and fed by their caregivers, respectively, when with diarrhea. No existing published literature exists to explain potential ethnic differences in household care practices for diarrhea in Malawi, suggesting the need for in-depth studies in order to understand socioeconomic and/or cultural factors that may support these practices at the household and community level. In conclusion, the study findings carry some important general pointers in terms of policy. Of high significance are the district influences on child morbidity management. In particular, the results suggest that in Malawi, Northern Region districts generally respond poorly to childhood diarrhea compared to those in Southern Region. The findings draw our attention to the need for concerted efforts on the part of government, but also for policy makers to pay attention to particular social and environmental conditions at district and regional level in designing policy responses to childhood diarrhea. The findings of this study also suggests that, given political will, households could be empowered to take more control over the management of diarrheal cases within their homes and communities as part of an effort to reduce under-five morbidity and mortality in Malawi.
0.6
References 0.4 0.2 0 -0.2 -0.4 -0.6
1
6
11
16
21
26
31
Mother's age in years Fig. 6. Nonlinear effects of child’s and mother’s age on childhood feeding.
We controlled for spatial dependence in the data where the fixed effects show the importance of paternal education, wealth index, breastfeeding, and ethnicity on oral rehydration while paternal education, marital status and ethnicity show significant influence on feeding. The effect of paternal education and wealth on oral rehydration in this study is generally consistent with studies that have shown that household socio-economic status is associated with child survival since it determines the amount of resources (such as food, good sanitation, and health care) available
Adebayo, S.B., Fahrmeir, L., 2005. Analysing child mortality in Nigeria with geoadditive discrete‐time survival models. Stat. Med. 24 (5), 709–728. Belitz, C., Brezger, A., Kneib, T., Lang, S., Umlauf, N., 2012. BayesX: Software for Bayesian Inference in Structured Additive Regression Models, Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. Bhandari, N., Bahl, R., Mazumdar, S., Martines, J., Black, R.E., Bhan, M.K., 2003. Effect of community-based promotion of exclusive breastfeeding on diarrhoeal illness and growth: a cluster randomised controlled trial. Lancet 361 (9367), 1418–1423. Boschi-Pinto, C., Velebit, L., Shibuya, K., 2008. Estimating child mortality due to diarrhoea in developing countries. Bull. World Health Organ. 86 (9), 710–717. Carter, R., Mendis, K.N., Roberts, D., 2000. Spatial targeting of interventions against malaria. Bull. World Health Organ. 78 (12), 1401–1411. Gayawan, E., Adebayo, S.B., 2013. A Bayesian semiparametric multilevel survival modelling of age at first birth in Nigeria. Demogr. Res. 28. Gwatkin, R., Rustin, S., Johnson, K., Pande, P., Wagastaff, A., 2003. Socio-economic Differences in Health, Nutrition and Population in India. HNP/Poverty Thematic Group, World Bank p. May, 2000. Kalipeni, E., 1993. Determinants of infant mortality in Malawi: a spatial perspective. Soc. Sci. Med. 37 (2), 183–198. Kandala, N., Madise, N., 2004. The Spatial Epidemiology of Childhood Diseases in Malawi and Zambia. African Population Studies, 19, 191–218. Kandala, N., Magadi, M.A., Madise, N.J., 2006. An investigation of district spatial variations of childhood diarrhoea and fever morbidity in Malawi. Soc. Sci. Med. 62 (5), 1138–1152. Kazembe, L.N., 2009. Modelling individual fertility levels in Malawian women: a spatial semiparametric regression model. Stat. Methods Appl. 18 (2), 237–255. Kazembe, L.N., 2007. Spatial modelling and risk factors of malaria incidence in northern Malawi. Acta Trop. 102 (2), 126–137. Malawi Government, 2008. Malawi Population and Housing Census Report. Statistical Office, Zomba.
94
D.O. Atari, P. Mkandawire / Health & Place 29 (2014) 84–94
Malawi Government, 2010. Malawi Demographic and Health Survey Report. Ministry of Health, Lilongwe. Malawi Government, 2012. Malawi Welfare Monitoring Report. Statistical Office, Zomba. Mbonye, A.K., 2004. Risk factors for diarrhoea and upper respiratory tract infections among children in a rural area of Uganda. J. Health Popul. Nutr. 22 (1), 52–58. Millard, A.V., 1994. A causal model of high rates of child mortality. Soc. Sci. Med. 38 (2), 253–268. National Statistical Office (NSO), ICF Macro, 2011. Zomba, Malawi and Calverton, Maryland. Posner, D.N., 2004. The political salience of cultural difference: why Chewas and Tumbukas are allies in Zambia and adversaries in Malawi. Am. Polit. Sci. Rev. 98 (4), 529–545.
Stockman, Lauren J., et al., 2007. Point-of-use water treatment and use among mothers in Malawi. Emerg. Infect. Dis. 13 (7), 1077. UNAIDS, 2009. AIDS Epidemic Update. WHO Regional Office Europe p. 2009 (December). UNICEF, 2012. State of the World’s Children 2012 UNICEF. Available at: 〈http:// www.unicef.org/sowc2012〉/. Wakefield, M.A., Loken, B., Hornik, R.C., 2010. Use of mass media campaigns to change health behaviour. Lancet 376 (9748), 1261–1271. Woldemicael, G., 2001. Diarrhoeal morbidity among young children in Eritrea: environmental and socioeconomic determinants. J. Health Popul. Nutr. 19 (2), 83–90.
Contents lists available at ScienceDirect
Health & Place journal homepage: www.elsevier.com/locate/healthplace
Spatial variation of management of childhood diarrhea in Malawi Dominic Odwa Atari a,n, Paul Mkandawire b a b
Department of Geography, Nipissing University, North Bay, ON, Canada The Institute of Interdisciplinary Studies, Carleton University, Ottawa, ON, Canada
art ic l e i nf o
a b s t r a c t
Article history: Received 29 January 2014 Received in revised form 24 May 2014 Accepted 19 June 2014
This study reports the spatial variability in household management of diarrhea among under-fives in Malawi. Using data from 2010 Malawi Demographic and Health Survey, we examined oral rehydration and feeding practices of mothers and caregivers of 3105 children with an episode of diarrhea by mapping district effect residual in geo-additive probit model and analyzing residual spatial effects in a Bayesian approach. The findings suggest that although diarrhea is relatively less prevalent in the Northern Region, this region lags behind in terms of adoption of appropriate practices for home-based management of diarrhea in children compared to the Central Regions and Southern Regions. A cluster of five predominantly rural districts in the eastern part of the Southern Region showed remarkably high level of household care for childhood diarrhea relative to the rest of the country. The fixed effects show the importance of breastfeeding, paternal education, wealth index, and ethnicity on oral rehydration, while paternal education, marital status, and ethnicity show significant influence on feeding for children with a diarrhea episode. The paper discusses the apparent inverse relationship between regional prevalence of diarrhea episodes and care-seeking practices for childhood diarrhea in Malawi, and makes relevant recommendations for policy. & 2014 Elsevier Ltd. All rights reserved.
Keywords: Diarrhea Children Spatial patterns Malawi Policy
1. Introduction Diarrhea is the passage of three or more loose stool or liquid per day. Diarrhea ranks amongst five most important causes of under-five morbidity and mortality in Malawi. In developing countries, diarrheal diseases account for an estimated 18–21% of all deaths in children younger than five years. This is equivalent to 1.5 million deaths per year (Boschi-Pinto et al., 2008). Nearly four in five of these deaths (78%) occur in Africa and South-East Asia. These two regions are also already excessively burdened by other infant and childhood diseases, including HIV/AIDS and malaria (Boschi-Pinto et al., 2008; UNAIDS, 2009). Although diarrhea can be dangerous as it can cause severe dehydration, simple household practices such as increasing the amount and frequency of drinks and food given to the child can replace the lost body fluids and salts and effectively treat the condition. Because treatment for diarrhea is relatively straightforward and inexpensive, health service providers, policy-makers, and researchers concerned with Integrated Management of Childhood Illnesses (IMCI) in poor countries seek to know about the spatial distribution of such child care practices in order to have a better understanding of how they can be promoted. Geographical
n
Corresponding author. Tel.: þ 1 705 474 3450x4196; fax: þ 1 705 474 1947. E-mail address: [email protected] (D.O. Atari).
http://dx.doi.org/10.1016/j.healthplace.2014.06.005 1353-8292/& 2014 Elsevier Ltd. All rights reserved.
location is thus an important factor in identifying clusters of the children who may be at elevated risk of diarrhea-related morbidity and mortality because of lack of awareness or poor adherence to these simple life-saving practices on the part of mothers or caregivers (Carter et al., 2000). Although there is expansive literature on childhood mortality and morbidity in Malawi in general, very few published studies have specifically focused on diarrhea. Available evidence however suggests that diarrhea-related under-five mortality and morbidity in Malawi is spatially highly uneven. A study conducted by Kandala et al. (2006), for instance, pointed to district-level socioeconomic and demographic (e.g. food security and population density) as important determinants of diarrhea-related childhood mortality. This study reinforced the findings of an earlier study conducted by Kalipeni (1993) which suggested that the district in which a child is born sets the context for risk for child morbidity and mortality. Kazembe (2007) also found dramatic variations in risk factors of co-morbidity related to fever, diarrhea and pneumonia in under-fives across regions in Malawi. This study also reported that age, habitual place of residence, nutrition status, and bed net use as important determinants, and urged more research to determine the precise linkages between childhood health and geographic location. In another study, Stockman et al. (2007) pointed to variations in access to safe and potable water as an important determinant of spatial differences in prevalence of under-five diarrhea. The authors found that rural poor women
D.O. Atari, P. Mkandawire / Health & Place 29 (2014) 84–94
are less likely to be aware of and to use simple diarrheal disease prevention techniques (e.g. point-of-use water treatment) relative to their urban counterparts. This study adds to this literature by examining both the suitability and adequacy of therapeutic responses of mothers and caregivers of children with known signs of diarrhea in Malawi. We specifically focus on whether there are significant geographic variations in childhood diarrhea management practices at the household level and, if so, the kind of factors that could explain such variations.
2. Context Malawi is located in the south-eastern region of the African continent (Fig. 1) with an estimated population of 14 million. Malawi is one of the poorest countries in the world with a per
85
capita income of US$383 and approximately 70% of Malawians live on less than one US$/day (Malawi Government, 2010). The majority of the population (80%) lives in rural areas and subsists on agriculture as smallholder famers. High levels of poverty are also linked to growing population pressure. Malawi was divided into three administrative regions after the country gained political independence from Britain in 1964; Northern Region, Central Region and Southern Region. The Northern Region comprises of five districts, Central Region has 9 districts, while the Southern Region has a total of 13 districts. The Southern Region is most populous, accounting for 45% of national population compared to 42% in the Central Region, and 13% in the Northern Region. High population growth rates relative to the fixed amount of landmass has meant that the population density in the Southern Region is nearly 3 times higher (185) than the Northern Region (63). The Central Region has a population density of 154 (Malawi Government, 2008).
Fig. 1. Study area.
86
D.O. Atari, P. Mkandawire / Health & Place 29 (2014) 84–94
Malawi has one of the highest under-five mortality rates in the world. According to the 2010 Malawi Demographic and Health Survey report, infant mortality is currently estimated at 66 per 1000 live births, down from 81 per 1000 live births in 2004. Similarly, a recent report from the United Nations named Malawi as one of the top 30 most dangerous places for children under-five, with an overall mortality rate of 92 deaths per 1000 live births in 2010 (in comparison, Canada and most European countries have rates between 2 and 6) (UNICEF, 2012). This is down from 145 per 1000 live births in 2004 (Malawi Government, 2010). These improvements in child survival are in part the result of policy of Integrated Management of Childhood Illnesses (IMCI) which has been implemented by the government in collaboration with UNICEF and WHO since 1999. A key component of IMCI is community integrated management of childhood illnesses which involves improving the ability of mothers and caregivers at the household and community levels to promptly recognize, treat and refer suspected cases of malaria, diarrhea, measles, acute and respiratory infections in under-fives. Despite apparent improvements, the current rate of under-five mortality is still excessively high. The 2010 MDHS report states that gender relations, birth order, mother’s education, and place of residence are important determinants of child mortality and morbidity in Malawi (National Statistical Office (NSO), ICF Macro, 2011). Access to essential health care is also low. For instance, only 80% of children are fully vaccinated, with those rural areas more likely to be vaccinated at 82% than those in urban areas at 76% (Malawi Government, 2010). Nearly one in every 10 children under the age of five present symptoms of acute respiratory infection in Malawi, but only seven out of every 10 of these children receive appropriate treatment within 24 h of onset of the signs. Similarly, prompt treatment for presumed malaria among under-fives remains low. Malawi Demographic and Health Survey (MDHS, 2010) shows that only 61% of under-five children presenting fever where taken to a health facility for treatment in the two weeks prior to the survey. There are considerable regional and district variations in factors related to the risk and management of diarrheal conditions in Malawi. For instance, levels of educational attainment vary dramatically across the three administrative regions. Due to early exposure to Scottish missionary education, the Northern Region’s historic advantage in functional literacy over the other regions continues into the present day. For instance, nearly 90% of men and 80% of women can read and write in the Northern Region, compared to 80% among men and 60% among women in the Southern Region. Beyond population densities and literacy rates, the three regions in Malawi also differ on a number of other relevant dimensions. Despite having high level of literacy, the Northern Region is politically weak and significantly underdeveloped compared to other regions. A recent welfare monitoring report revealed significant regional and district level variations in access to social amenities, as measured by a number of variables including ease of access to public transportation or hospitals. The report indicated, for instance, that 22% of household in the Northern Region live within less than an hour of the nearest public transport port compared to 33% in the Southern Region (Malawi Government, 2012). In the eastern districts of the Sothern Region, such as Thyolo, access to public transportation is much higher, with over 50% of the population living within less than 1 h of a public transport port. Similarly, only 68% of households in the Northern Region live within a 15 min-walk to the nearest source of drinking water, compared to 77% in the Southern Region. In eastern districts of the Southern Region, such as Phalombe, more that 82% of the households are located within a 15 min-walk to the nearest source of water. In terms of household practices for prevention of diarrheal diseases, such as boiling drinking water,
there is a clear north–south gradient with only 15% of households in the Northern Region reporting boiling water before drinking, compared to 20% in the Central Region and 25% in the Southern Region. In some districts in the eastern part of the Southern Region, such as Thyolo, the proportion of households reporting boiling drinking water is high as 36%.
3. Data and methods The study applied Bayesian hierarchical model and geostatistical techniques with location (district), and nonlinear metrical (mother’s and child’s age) attributes and other information to gain a better understanding of the spatial variation in the management of diarrhea in Malawi. The study uses the 2010 MDHS, the most recent nationally representative dataset of demographic and health collected by Malawi Government in collaboration with MEASURE (National Statistical Office (NSO), ICF Macro, 2011). A more detailed explanation of the data collection and methodological procedures followed by the 2010 MDHS are explained elsewhere (National Statistical Office (NSO), ICF Macro, 2011), but we highlight that the 2010 MDHS data were collected using a twostage stratified sampling design to provide national, regional and district estimates of health and demographic indicators. In the first stage, 849 enumeration areas, or clusters, were selected as primary sampling units, stratified by urban/rural status using sampling probability proportional to the population of the selected cluster. In the second stage, a fixed number of households were randomly selected from each cluster. A total of 13,220 women of reproductive age range (15–49 years) were interviewed using an interviewer-administered questionnaire which included questions capturing information on birth histories. A sub-sample of 3105 records was drawn involving all cases where a mother or caregiver reported an episode of child under-five diarrhea in the two weeks prior to the survey for use for the present study. The 2010 MDHS asked respondents “Has your child had diarrhea in the last 2 weeks?”. If the mother answers yes, a follow-up question was asked as to whether she provided their sick child with “less than usual”, “the same as usual”, or “more than usual” amount of fluids (oral rehydration) and food (feeding). Theoretically relevant predictors including individual variables (child’s sex, birth order, age, and breastfeeding condition), family (mother’s age, parental education, number of children under five, marital status, wealth index, and household size), and community (place of residence, ethnicity, and district of residents) characteristics were examined. Birth order was categorized into ‘1st’, ‘2nd and 3rd’, ‘4th and 5th’, and ‘6 or higher’, and breastfeeding status was dichotomized into ‘Yes’ for mothers who were breastfeeding at the time of the study and ‘No’ for those who were not. Maternal and paternal educational attainment was categorized as ‘no education’, ‘primary education’, and ‘secondary or higher education’. The number of children under the age of five in the household was categorized as ‘1 child’, ‘2 children’, and ‘3 or more children’. Furthermore, marital status was dichotomized into two categories; ‘married or living together’ and ‘Single’ mothers (never married, widowed, divorced, and separated). The 2010 MDHS also has a wealth index which captures the long-term standard of living based on household’s ownership of consumer goods; dwelling characteristics; type of drinking water source; toilet facilities; and other relevant characteristics related to a household socioeconomic status (National Statistical Office (NSO), ICF Macro, 2011; Gwatkin et al., 2003). Wealth quintiles included ‘Poorest’, ‘Poor’, ‘Middle’, ‘Rich’ and ‘Richest’. Household size was categorized as ‘small’ (if there were fewer than 5 members), ‘medium’ (if there were 5–9 members), and ‘large’ (more than 10 members). Place of residence was dichotomized into
D.O. Atari, P. Mkandawire / Health & Place 29 (2014) 84–94
87
Table 1 Distribution and univariate ordinal regression analysis of factors in childhood diarrheal management in Malawi (MDHS, 2010). Individual characteristics
Sex of child Male Femaleb Birth order 1st 2nd and 3rd 4th and 5th 6þ b Child’s age in monthsa o 6 months 6–11 12–23 24–35 36 þ Monthsb Breastfeeding No Yesb Family characteristics Mother’s age in yearsa o21 Years 21–24 25–29 30–34 35–39 40 þYearsb Maternal education No education Primary education Sec. education or higherb Paternal education No education Primary education Sec. education or higherb Number of under fives 1 child 2 children 3 þ childrenb Marital Status Married/living together Single mothersb Wealth index Poorest Poor Middle Richer Richestb Household size Small ( o 5 members) Medium (5–9 members) High ( 49 members)b Community characteristics Place of residence Urban Ruralb Ethnicity Chewa Tumbuka Lomwe Tonga Yao Sena Nkonde Ngoni Amanganja/Anyanja Others
Count (%)
Rehydration
Feeding
Estimate
95% CI
Estimate
95% CI
1575 (50.7) 1530 (49.3)
0.03
0.05
0.11
0.01
0.10
0.08
657 (21.2) 1134 (36.5) 718 (23.1) 596 (19.2) 21.4 713.6 147 (4.7) 750 (24.2) 1104 (35.6) 582 (18.7) 522 (16.8)
0.05 0.04 0.01
0.08 0.07 0.14
0.18 0.16 0.11
0.12 0.13 0.01
0.02 0.01 0.14
0.25 0.25 0.13
0.01 0.08 0.02 0.01
0.20 0.21 0.10 0.14
0.22 0.05 0.13 0.12
0.65 0.20 0.23 0.02
0.90 0.33 0.35 0.15
0.40 0.07 0.10 0.13
0.01
0.11
0.08
0.16
0.06
0.25
0.03 0.03 0.05 0.01 0.03
0.22 0.15 0.13 0.18 0.23
0.16 0.20 0.22 0.19 0.17
0.10 0.06 0.02 0.06 0.15
0.30 0.13 0.20 0.26 0.36
0.10 0.24 0.17 0.13 0.07
484 (15.6) 2225 (71.7) 396 (12.8)
0.16 0.03
0.31 0.16
0.01 0.09
0.10 0.05
0.26 0.18
0.07 0.08
406 (13.1) 1924 (62) 775 (25)
0.16 0.16
0.29 0.26
0.02 0.07
0.10 0.12
0.24 0.23
0.05 0.02
1144 (36.8) 1492 (48.1) 469 (15.1)
0.08 0.01
0.05 0.11
0.20 0.13
0.05 0.05
0.08 0.18
0.18 0.08
2678 (86.2) 427 (13.8)
0.06
0.06
0.18
0.05
0.08
0.17
0.21 0.22 0.07 0.07
0.35 0.36 0.21 0.22
0.06 0.08 0.07 0.08
0.03 0.13 0.03 0.03
0.18 0.28 0.12 0.18
0.12 0.03 0.18 0.13
1018 (32.8) 1883 (60.6) 204 (6.6)
0.06 0.01
0.11 0.16
0.23 0.17
0.06 0.06
0.12 0.24
0.24 0.12
290 (9.3) 2815 (90.7)
0.15
0.01
0.28
0.11
0.03
0.26
1080 (34.8) 319 (10.3) 490 (15.8) 49 (1.6) 327 (10.5) 151 (4.9) 46 (1.5) 400 (12.9) 145 (4.7) 98 (3.2)
0.36 0.24 0.65 0.17 0.33 0.41 0.18 0.68 0.45
0.11 0.03 0.39 0.23 0.06 0.11 0.62 0.42 0.15
0.60 0.51 0.90 0.58 0.60 0.71 0.25 0.95 0.75
0.18 0.30 0.38 0.22 0.38 0.56 0.63 0.32 0.25
0.09 0.01 0.11 0.21 0.09 0.24 1.17 0.04 0.08
0.45 0.59 0.66 0.65 0.67 0.87 0.08 0.60 0.58
786 (25.3) 2319 (74.7) 27.3 7 6.8 487 (15.7) 782 (25.2) 816 (26.3) 500 (16.1) 315 (10.1) 205 (6.6)
767 (24.7) 695 (22.4) 723 (23.3) 531 (17.1) 389 (12.5)
Rehydration and feeding are ordered ordinal variables (1 ¼less than usual, 2 ¼same as usual, 3¼ more than usual). a b
Mean 7 Standard deviation. Reference category.
‘Rural’ and ‘Urban’. Malawi is an ethnically diverse country comprising of 10 major tribes; Chewa, Tumbuka, Lomwe, Tonga, Yao, Sena, Nkonde, Ngoni, Mang’anja, and Nyanja. This study
therefore also controlled for ethnicity. The distributions of variables analyzed in the management of childhood diarrhea in Malawi are outlined in Table 1.
88
D.O. Atari, P. Mkandawire / Health & Place 29 (2014) 84–94
Fig. 2. Spatial distribution of observed diarrheal prevalence by regions (left) and districts (rights) in Malawi.
3.1. Statistical analysis The dependent variable is defined as yi ¼1 if child i had received oral rehydration or feeding that was less than usual, 2 same as usual, or 3 more than usual. The commonly adopted model for the analysis of this type of data is the probit or logistic model, and the standard measure of effects is the estimated mean or odds ratio (OR) (Mbonye, 2004; Woldemicael, 2001). However, because of the spatial nature of our data and the presence of nonlinear effects for some covariates, the assumption of a linear relationship between predictors and the dependent variable may not hold leading to biased parameter estimates. Hence, a flexible
Geo-additive logistic model was used (on the probability of a child receiving sufficient rehydration or feeding) to determine the socioeconomic and demographic variables associated with management of diarrhea while simultaneously controlling for spatial dependency in the data and possible nonlinear effects of the independent variables. Both oral rehydration and feeding were examined as ordinal dependent variables. Due to the ordered ordinal nature of the dependent variables (oral rehydration and feeding), the following cumulative probit model was used:
yi
(r)
( f1(xi1)
+ fp(xip) + f spat (si
i
it
ð1Þ
D.O. Atari, P. Mkandawire / Health & Place 29 (2014) 84–94
Table 2 Regions and districts of Malawi. Region and districts Region Northern
Central
Southern
Count (%) District Chitipa Karonga Nkhatabay Rumphi Mzimba
N ¼ 3105 77 (2.5) 82 (2.6) 36 (1.2) 109 (3.5) 124 (4.0)
Total Kasungu Nkhota Kota Ntchisi Dowa Salima Lilongwe Mchinji Dedza Ntcheu
428 (13.8) 222 (7.1) 134 (4.3) 136 (4.4) 128 (4.1) 126 (4.1) 162 (5.2) 122 (3.9) 131 (4.2) 92 (3.0)
Total Mangochi Machinga Zomba Chiradzulu Blantyre Mwanza Thyolo Mulanje Phalombe Chikwawa Nsanje Balaka Neno
1253 (40.4) 83 (2.7) 128 (4.1) 129 (4.2) 89 (2.9) 120 (3.9) 90 (2.9) 96 (3.1) 119 (3.8) 195 (6.3) 70 (2.3) 115 (3.7) 94 (3.0) 96 (3.1)
Total
1424 (45.9)
where Φ is the standard normal cumulative distribution function, yi A {1, 2, 3} denotes the rehydration or feeding which child i received, and f1, …, fp are nonlinear smooth effects of covariates (e. g. child’s and mother’s age) while fspat is the effect of district si A {1, …, S} where child i lives. The standard measure of effect in the probit model is the mean. However, because we used the Bayesian approach that relies on prior assumptions to make posterior inference, we used the posterior mean instead. To account for possible departures from the assumed distribution, 95% confidence intervals (CIs) for the posterior odd ratios (ORs) and probability maps (the equivalent of CIs for the spatial effects) are calculated using robust standard errors estimated via Markov Chain Monte Carlo (MCMC) simulation techniques. The analysis was carried out using BayesX-version 2.1 (Belitz et al., 2012), a software for Bayesian inference based on MCMC simulation techniques. The posterior distribution is intractable, so MCMC algorithms are used to generate samples from this prior distribution, which allows estimation and inference for the parameters. For each of the rehydration and feeding models considered in this study, 12,000 iterations were carried out with a burn-in period of 2000. We also investigated sensitivity to choice of priors by varying the values of hyper-parameters a and b and the results are stable for a ¼b¼0.001. This is consistent with what other authors have experienced (e.g. Adebayo and Fahrmeir, 2005; Kazembe, 2009; Gayawan and Adebayo, 2013).
4. Results The results show that overall 45.6% of under-five children received less than usual amount of oral rehydration while 33.6% and 20.7% received the same or more than usual amount of fluids
89
during a diarrhea episode, respectively. In terms of feeding, 64.2% of under-fives were fed less than usual amount of food during a diarrhea episode compared to only 28.9% and 7.2% who were provided with the same or more than usual amount of feeding, respectively. Table 1 shows the distribution and univariate ordinal regression analysis of factors related to childhood management of diarrhea by caregivers in Malawi. The sample had approximately equivalent number of male (50.7%) and female children. More than 21% of the children were first order births while 19% were at least the 6th order births. The mean age of the children was 21.4713.6 months. The average age of mothers of children was 27.376.8 years. Nearly 37% of mothers had 1 child under the age of five, while 48% and 15% of mother had two, and three or more underfive children, respectively. Approximately 25% of children were still breastfeeding at the time of the survey. With regards to education, only 13% of mothers completed secondary school or higher. The univariate ordinal regression results (Table 1) indicate that mother’s age, paternal education, wealth index, place of residence, and ethnicity are significant determinants of oral rehydration at the household level, while birth order, child’s age, breastfeeding condition, paternal education, wealth index, and ethnicity show significant influence on feeding. Fig. 2 shows spatial variations in the prevalence of diarrhea in Malawi. The results indicate clear disparities. Regionally, the Northern Region shows the lowest prevalence, followed by the Southern Region and the Central Regions (Fig. 2a). At the district level, highest prevalence was found in the Central Region district of Kasungu and the Southern Region district of Phalombe (Fig. 2b). Table 2 indicates the distribution of the study subjects amount the regions and districts of Malawi. The results for oral rehydration, as presented in Fig. 3, suggest considerable spatial auto-correlation in the underlying posterior means. The left panel reveals a cluster of high-risk practices by mothers who provide less than usual amount of fluids to their children during a diarrhea episode in the Northern Region districts of Chitipa, Karonga, Rumphi, and Mzimba compared to mothers in the southern districts of Zomba, Phalombe, Mulanje, Chiradzulu, and Thyolo who were more likely to provide more than usual rehydration to their children in case of diarrhea. The results of the nonlinear effect of child’s age (Fig. 4a) on oral rehydration suggest that there was an increasing potential for mothers to provide more oral rehydration to their children when they were younger than 26 months after which it levels off. We show estimated posterior means together with 95% CIs. Although there is a general tendency for mothers to rehydrate their children with increasing age, the influence of mother’s age on child rehydration was also nonlinear (Fig. 4b). With regard to the fixed parameters, Table 3 shows that children who were not being breastfed were also less likely to be orally rehydrated more than usual compared to those who were breastfed during diarrhea episodes. Children whose fathers completed only primary education were less likely to receive more than usual amounts of oral rehydration compared to those whose fathers had university training. With respect to poverty, women in the poorest or poor household were reportedly less likely to provide higher amounts of oral rehydration to their kids with diarrhea compared to those in the richest households. Children of the Ngoni ethnic group were more likely to get more than usual rehydration treatment when they have diarrhea than children of other ethnic groups in Malawi. We did not find any statistically significant associations between the likelihood of providing more than usual amount of oral rehydration and child’s sex, birth order, maternal education, number of children under-five, marital status, household size, and place of residence.
90
D.O. Atari, P. Mkandawire / Health & Place 29 (2014) 84–94
Black – significant positive effect Grey – Not significant White – significant negative effect
Fig. 3. Residual spatial districts effects (left) and 95% posterior probability map (right) of childhood rehydration during diarrheal episodes in Malawi.
Similar to the spatial distribution of oral rehydration, Fig. 5 shows that children in the Northern Region districts of Chitipa, Karonga, Rumphi, and Mzimba and Central Region district of Nkhota-Kota were less likely to be fed more than usual amount of food compared to children in the Southern Region districts of Mulanje, Chiradzulu, Thyolo, and Nsanje who were more likely to receive more than usual amount of food when they have diarrheal episodes. Children in the other districts showed no significant differences in their tendency to be fed more than usual amount of food during diarrhea. Table 4 shows the fixed parameters with paternal education showing significant influence on children feeding. Married mothers or those living together with their partners were more likely to feed their sick child with more than usual amount of food. Children of the Tumbuka ethnic group were more likely to be fed more than usual amount of food when they have diarrhea compared to children of other ethnic groups in Malawi. Fig. 6a shows the nonlinear effect of child’s age on feeding. There is an increasing tendency for children under 30 months old to be fed more than usual amount of food when they have
diarrhea after which the age effect levels off. Overall, mother’s age indicate nonlinear effects on child’s diarrheal management (Fig. 6b).
5. Discussion and conclusions This study has shown significant district-specific geographical variation in the management of childhood diarrhea at the household level in Malawi. When compared, the north–south pattern of diarrhea, prevalence remains higher in the Northern Region relative to the Southern Region (Fig. 2a). The posterior mean estimates of the residual smooth spatial district effects are shown in the left panel of maps of Figs. 3–5. In addition, posterior probability maps (right panel of Figs. 3–5) indicate significance of the spatial effects (shades of white colored¼negatively significant, shades of black colored¼positively significant, and shades of grey colored ¼not significant). Note that the residual spatial effects are centered about zero, i.e. the average of all districts is zero.
D.O. Atari, P. Mkandawire / Health & Place 29 (2014) 84–94
91
Table 3 District posterior mean estimates of the fixed effect parameters for childhood rehydration during diarrheal cases in Malawi.
0.3 0.2
Variable 0.1 0 -0.1 -0.2 -0.3 -0.4
1
6
11
16
21
26
31
36
41
46
51
56
Child's age in months
0.6 0.4 0.2 0
Constant Breastfeeding No Yesa Paternal education No education Primary education Secondary education or highera Wealth index Poorest Poor Middle Richer Richesta Ethnicity Chewa Tumbuka Lomwe Tonga Yao Sena Nkonde Ngoni Amanganja/Anyanja Othersa
Mean
Std. error
95% CI
0.02
0.21
0.42
0.40
0.13
0.06
0.25
0.00
0.11 0.13
0.08 0.05
0.26 0.24
0.04 0.02
0.19 0.22 0.08 0.05
0.09 0.08 0.08 0.08
0.36 0.39 0.25 0.21
0.02 0.06 0.08 0.11
0.28 0.31 0.22 0.21 0.13 0.17 0.11 0.48 0.10
0.16 0.16 0.17 0.24 0.17 0.21 0.26 0.17 0.18
0.02 0.01 0.10 0.25 0.20 0.22 0.40 0.16 0.25
0.60 0.64 0.55 0.66 0.47 0.58 0.62 0.81 0.46
-0.2 a
Reference category.
-0.4 -0.6
15
20
25
30
35
40
45
Mother's age in years Fig. 4. Nonlinear effects of child’s and mother’s age on childhood rehydration.
Over and above the impact of the fixed effects, there appear to be widespread negative influences on oral rehydration and feeding of children with diarrhea in the Northern Region districts. Remarkably, despite high levels of functional literacy compared to national average, children with diarrheal episodes in the Northern Region received less than recommended amounts of oral rehydration and feeding relative to children in the Central Region and Southern Region districts. Poor management of diarrhea in the Northern Region districts however exists within an epidemiological context of widespread low prevalence of the disease compared to the other regions (Table 2) (National Statistical Office (NSO), ICF Macro, 2011). This means that episodes of childhood diarrhea in this region are relatively low relative to the other regions, yet when the disease occurs, caregivers may not have appropriate level of knowledge and skills to properly manage it at home. This may be due to a number of reasons. First, low prevalence of diarrhea in the Northern Region may be diverting policy attention towards those areas where the disease is most prevalent. Second, in a context where resources are limited, prioritizing high prevalence areas is understandable. Third, the region’s low population density, with most of its population dispersed over a rugged topography with poor feeder road network has meant that efficient delivery of health programs is always a challenge. For instance, districts such as Chitipa, Rumphi, and Nkhata-Bay are mountainous with large swaths of hard-to-reach populations, in some cases only reachable by boat. Finally, widespread poor health-seeking behaviour for diarrhea in the Northern Region may be linked to historical factors where, lacking political power, the districts in the region find it relatively difficult to attract
government resources and policy attention as compared to the districts from the other regions. Unlike the Northern Region, Central Region districts show no significant effect in childhood diarrhea management. This intermediate status may correspond to a range of political, socioeconomic and environmental factors that characterize the region. For instance, while relatively high population density and mobility, and rapid urbanization may foster the spread of diarrhea, being the region that houses the capital city has always meant that the Central Region has always enjoyed a level of access to government resources and policy that is unknown to the other regions (Posner, 2004). In addition to political clout, the Central Region’s relatively less challenging topography may make it attractive for organizations to provide health education and delivery health services even in rural remote areas as evident by the high number of NGO in most districts in the region. Thus, geographic factors in the region may interact with social and political factors to keep diarrhea in check, as opposed to the Northern Region. High level of population density, increased population mobility, and sprawled conditions in small towns and estates in the Southern Region are well-known risk factors for the spread of diarrhea (Malawi Government, 2012). At the same time, large swaths of densely populated areas with people living in close proximity in the agricultural frontiers and rural growth centers of the region may facilitate rapid diffusion of health messages and practices. In addition, high population density can also lead to dense social networks which may have the secondary effect of creating a more supportive environment for adoption and sustainment of positive household health-seeking behavior (Wakefield et al., 2010). Remarkable improvements in the care of diarrhea exhibited by a cluster of five Southern Region districts of Zomba, Mulanje, Chiradzulu, Thyolo, and Phalombe should also be understood within the context of the growing political attention that these districts have garnered over the past two decades and an increase in resource flows to those districts that has accompanied the attention. These five districts account for about 20% of Malawi’s
92
D.O. Atari, P. Mkandawire / Health & Place 29 (2014) 84–94
Black – significant positive effect Grey – Not significant White – significant negative effect
Fig. 5. Residual spatial districts effects (left) and 95% posterior probability map (right) of childhood feeding during diarrheal episodes in Malawi.
population and constitute the strongest political base for all the three state presidents that have ruled Malawi since multiparty democracy in 1994. Concerted government policy in these districts over the decades, coupled with increased service delivery by NGOs and other civil society organizations may explain the remarkably high level of positive health behavior in household management of diarrhea, despite a widespread low level of literacy. The fact that these five largely rural districts outperformed even Blantyre, the country’s commercial capital city, gives hope that the apparently onerous task of reducing diarrhea-related morbidity and mortality in Malawi is not an impossible project. The example demonstrated by the five districts suggests that, given the right conditions, communities and households can be empowered to successfully prevent and treat cases of diarrhea in their children and therefore stop unnecessary deaths from occurring. For instance, the government could potentially capitalize on already high rate of functional literacy in the Northern Region to intensify health literacy, thereby enhancing the ability of mothers in this region to care for their children during an episode of diarrhea. In addition, investment in transport and communication
infrastructure can reduce logistical challenges that currently undermine effective health service delivery in various parts of the country, especially the Northern Region. This can help to expand coverage and facilitate scale-up of health services to hard-to-reach communities, especially those in hard-to-reach areas of the Northern Region which may be ill-equipped to deal with diarrhea due to being unfamiliar with the disease or lack of prior exposure to relevant health education and skills. In terms of individual-level parameters, the results indicate that the effects of child’s and mother’s age on the amount of oral rehydration and feeding were predominantly nonlinear (Figs. 4–6). The analysis shows that older children are more likely to be provided with adequate amounts of oral rehydration and feeding compared to their younger counterparts. The general tendency for the frequency of child feeding to increase during an episode diarrhea with the mother’s age may in part explain a known trend of increasing childhood morbidity among young mothers who are often not only economically disadvantaged, but also less experienced in terms of child caring relative to older mothers Kandala and Madise, 2004)
D.O. Atari, P. Mkandawire / Health & Place 29 (2014) 84–94
Table 4 District posterior mean estimates of the fixed effect parameters for childhood feeding during diarrheal cases in Malawi. Variable
Mean
Std. error
95% CI
Constant Paternal education No education Primary education Secondary education or highern Marital status Married/living together Single mothersn Ethnicity Chewa Tumbuka Lomwe Tonga Yao Sena Nkonde Ngoni Amanganja/Anyanja Othersn
0.53
0.22
0.95
0.11
0.06 0.12
0.08 0.06
0.22 0.24
0.10 0.00
0.15
0.06
0.04
0.26
0.15 0.40 0.10 0.21 0.20 0.17 0.50 0.20 0.01
0.17 0.17 0.18 0.25 0.18 0.22 0.32 0.18 0.20
0.18 0.06 0.25 0.29 0.16 0.26 1.13 0.13 0.37
0.48 0.74 0.45 0.71 0.56 0.61 0.11 0.55 0.39
0.6 0.4 0.2 0 -0.2 -0.4 -0.6 -0.8 -1 -1.2
1
6
11
16
21
26
31
36
41
46
51
56
Child's age in months
93
to children (Millard, 1994). For instance, while examining child morbidity in Malawi and Zambia, Kandala and Madise (2004) found that childhood morbidity was lower among educated women and lower morbidity was also reported in households with large number of adult members. In addition, it is not surprising that breastfeeding emerged as an important factor for oral rehydration. IMCI policy guidelines stipulate that mothers of children with diarrhea should be encouraged to breastfeed their children more frequently as part of oral rehydration. One of the major reasons why exclusive breastfeeding is encouraged in developing countries relates to the need to reduce the risk of diarrhea in infants. This means that mothers who practice breastfeeding are more likely to know about its benefits in relation to diarrhea case management in children (Bhandari et al., 2003). There are also ethnic differences in both rehydration and feeding for diarrheal management in Malawi. For example, children from the Ngoni and Tumbuka ethnic groups were more likely to be sufficiently rehydrated and fed by their caregivers, respectively, when with diarrhea. No existing published literature exists to explain potential ethnic differences in household care practices for diarrhea in Malawi, suggesting the need for in-depth studies in order to understand socioeconomic and/or cultural factors that may support these practices at the household and community level. In conclusion, the study findings carry some important general pointers in terms of policy. Of high significance are the district influences on child morbidity management. In particular, the results suggest that in Malawi, Northern Region districts generally respond poorly to childhood diarrhea compared to those in Southern Region. The findings draw our attention to the need for concerted efforts on the part of government, but also for policy makers to pay attention to particular social and environmental conditions at district and regional level in designing policy responses to childhood diarrhea. The findings of this study also suggests that, given political will, households could be empowered to take more control over the management of diarrheal cases within their homes and communities as part of an effort to reduce under-five morbidity and mortality in Malawi.
0.6
References 0.4 0.2 0 -0.2 -0.4 -0.6
1
6
11
16
21
26
31
Mother's age in years Fig. 6. Nonlinear effects of child’s and mother’s age on childhood feeding.
We controlled for spatial dependence in the data where the fixed effects show the importance of paternal education, wealth index, breastfeeding, and ethnicity on oral rehydration while paternal education, marital status and ethnicity show significant influence on feeding. The effect of paternal education and wealth on oral rehydration in this study is generally consistent with studies that have shown that household socio-economic status is associated with child survival since it determines the amount of resources (such as food, good sanitation, and health care) available
Adebayo, S.B., Fahrmeir, L., 2005. Analysing child mortality in Nigeria with geoadditive discrete‐time survival models. Stat. Med. 24 (5), 709–728. Belitz, C., Brezger, A., Kneib, T., Lang, S., Umlauf, N., 2012. BayesX: Software for Bayesian Inference in Structured Additive Regression Models, Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. Bhandari, N., Bahl, R., Mazumdar, S., Martines, J., Black, R.E., Bhan, M.K., 2003. Effect of community-based promotion of exclusive breastfeeding on diarrhoeal illness and growth: a cluster randomised controlled trial. Lancet 361 (9367), 1418–1423. Boschi-Pinto, C., Velebit, L., Shibuya, K., 2008. Estimating child mortality due to diarrhoea in developing countries. Bull. World Health Organ. 86 (9), 710–717. Carter, R., Mendis, K.N., Roberts, D., 2000. Spatial targeting of interventions against malaria. Bull. World Health Organ. 78 (12), 1401–1411. Gayawan, E., Adebayo, S.B., 2013. A Bayesian semiparametric multilevel survival modelling of age at first birth in Nigeria. Demogr. Res. 28. Gwatkin, R., Rustin, S., Johnson, K., Pande, P., Wagastaff, A., 2003. Socio-economic Differences in Health, Nutrition and Population in India. HNP/Poverty Thematic Group, World Bank p. May, 2000. Kalipeni, E., 1993. Determinants of infant mortality in Malawi: a spatial perspective. Soc. Sci. Med. 37 (2), 183–198. Kandala, N., Madise, N., 2004. The Spatial Epidemiology of Childhood Diseases in Malawi and Zambia. African Population Studies, 19, 191–218. Kandala, N., Magadi, M.A., Madise, N.J., 2006. An investigation of district spatial variations of childhood diarrhoea and fever morbidity in Malawi. Soc. Sci. Med. 62 (5), 1138–1152. Kazembe, L.N., 2009. Modelling individual fertility levels in Malawian women: a spatial semiparametric regression model. Stat. Methods Appl. 18 (2), 237–255. Kazembe, L.N., 2007. Spatial modelling and risk factors of malaria incidence in northern Malawi. Acta Trop. 102 (2), 126–137. Malawi Government, 2008. Malawi Population and Housing Census Report. Statistical Office, Zomba.
94
D.O. Atari, P. Mkandawire / Health & Place 29 (2014) 84–94
Malawi Government, 2010. Malawi Demographic and Health Survey Report. Ministry of Health, Lilongwe. Malawi Government, 2012. Malawi Welfare Monitoring Report. Statistical Office, Zomba. Mbonye, A.K., 2004. Risk factors for diarrhoea and upper respiratory tract infections among children in a rural area of Uganda. J. Health Popul. Nutr. 22 (1), 52–58. Millard, A.V., 1994. A causal model of high rates of child mortality. Soc. Sci. Med. 38 (2), 253–268. National Statistical Office (NSO), ICF Macro, 2011. Zomba, Malawi and Calverton, Maryland. Posner, D.N., 2004. The political salience of cultural difference: why Chewas and Tumbukas are allies in Zambia and adversaries in Malawi. Am. Polit. Sci. Rev. 98 (4), 529–545.
Stockman, Lauren J., et al., 2007. Point-of-use water treatment and use among mothers in Malawi. Emerg. Infect. Dis. 13 (7), 1077. UNAIDS, 2009. AIDS Epidemic Update. WHO Regional Office Europe p. 2009 (December). UNICEF, 2012. State of the World’s Children 2012 UNICEF. Available at: 〈http:// www.unicef.org/sowc2012〉/. Wakefield, M.A., Loken, B., Hornik, R.C., 2010. Use of mass media campaigns to change health behaviour. Lancet 376 (9748), 1261–1271. Woldemicael, G., 2001. Diarrhoeal morbidity among young children in Eritrea: environmental and socioeconomic determinants. J. Health Popul. Nutr. 19 (2), 83–90.
