Harmful practices in the management of childhood diarrhea in low- and middle-income countries: a systematic review

Background Harmful practices in the management of childhood diarrhea are associated with negative health outcomes, and conflict with WHO treatment guidelines. These practices include restriction of fluids, breast milk and/or food intake during diarrhea episodes, and incorrect use of modern medicines. We conducted a systematic review of English-language literature published since 1990 to assess the documented prevalence of these four harmful practices, and beliefs, motivations, and contextual factors associated with harmful practices in low- and middle-income countries. Methods We electronically searched PubMed, Embase, Ovid Global Health, and the WHO Global Health Library. Publications reporting the prevalence or substantive findings on beliefs, motivations, or context related to at least one of the four harmful practices were included, regardless of study design or representativeness of the sample population. Results Of the 114 articles included in the review, 79 reported the prevalence of at least one harmful practice and 35 studies reported on beliefs, motivations, or context for harmful practices. Most studies relied on sub-national population samples and many were limited to small sample sizes. Study design, study population, and definition of harmful practices varied across studies. Reported prevalence of harmful practices varied greatly across study populations, and we were unable to identify clearly defined patterns across regions, countries, or time periods. Caregivers reported that diarrhea management practices were based on the advice of others (health workers, relatives, community members), as well as their own observations or understanding of the efficacy of certain treatments for diarrhea. Others reported following traditionally held beliefs on the causes and cures for specific diarrheal diseases. Conclusions Available evidence suggests that harmful practices in diarrhea treatment are common in some countries with a high burden of diarrhea-related mortality. These practices can reduce correct management of diarrheal disease in children and result in treatment failure, sustained nutritional deficits, and increased diarrhea mortality. The lack of consistency in sampling, measurement, and reporting identified in this literature review highlights the need to document harmful practices using standard methods of measurement and reporting for the continued reduction of diarrhea mortality. Electronic supplementary material The online version of this article (doi:10.1186/s12889-015-2127-1) contains supplementary material, which is available to authorized users.


Background
Diarrheal disease is a leading cause of mortality in children under five, resulting in around 750,000 deaths each year [1]. The WHO recommends first line management of diarrhea in children under five with continued feeding, increased fluids, and supplemental zinc for 10-14 days to prevent dehydration. In addition, the WHO guidelines state that children exhibiting non-severe dehydration should "receive oral rehydration therapy (ORT) with ORS solution in a health facility". Antimicrobials are recommended only for the treatment of bloody diarrhea or suspected cholera with severe dehydration [2]. The full guidelines, which have evolved over time, are available at http://www.who.int/entity/maternal_child_adolescent/ documents/9241593180/en/index.html.
For decades, health initiatives have targeted the expansion of ORS and ORT, including the UNICEF Growth Monitoring, Oral Rehydration, Breastfeeding and Immunization (GOBI) initiative, the USAID/CDC Africa Child Survival Initiative -Combatting Childhood Communicable Diseases (ACSI-CCCD), and the WHO Integrated Management of Childhood Illness (IMCI) initiative. Despite these efforts, a shift in global attention away from diarrhea management seems likely to have contributed to slowingand even reversalsin progress toward full coverage for ORT [3,4].
Many fewer programs have specifically targeted nonadherence to other recommended diarrhea management practices, such as the restriction of fluids, breast milk and/ or food intake during diarrhea episodes, and incorrect use of modern medicines. All four of these practices are associated with negative outcomes and conflict with WHO treatment guidelines. Curtailment of fluids and restriction of feeding during diarrhea can increase the risk of dehydration, reduce nutritional intake, and potentially inhibit child growth and development. The use of antibiotics and other medications is appropriate only in the treatment of cholera or dysenteric diarrhea in children. Antidiarrheal drugs and some antiemetics not only have no benefit in diarrhea treatment, but may also cause serious, even lifethreatening side effects in children [2]. We have referred to these as "harmful practices" from this point forward, understanding that under some circumstances these practices may not be detrimental.
This review summarizes existing literature on harmful practices in diarrhea case management in children under five years of age, including fluid and breastfeeding curtailment, food restriction, and inappropriate use of medications for diarrhea management in children in low-and middle-income countries. The primary objectives of the review are to: Determine the documented prevalence of these four harmful practices across low-and middle-income populations, as reported in various studies since 1990; Describe how these practices have been examined and reported on previously; Explore beliefs, motivations, and contextual factors associated with harmful practices as reported through both quantitative and qualitative studies; and Highlight associations between these harmful practices and other characteristics of the episode, child, caregiver, and household.
Findings from this review will identify critical next steps to address harmful practices in diarrhea management and ultimately improve child survival.

Methods
We searched PubMed, Embase, Ovid Global Health, and the WHO Global Health Library in September 2013. Papers were identified that included variations on the combination of the following terms within the publication's title or abstract or as a keyword: 1) diarrhea; 2) low-and middle-income country; and one or more terms related to 3) a harmful practice or general management of diarrhea. Search terms were developed in PubMed (see Additional file 1) and translated for the three other databases. Publications were restricted to English-language articles published after 1990.
Quantitative articles were included if the paper reported the prevalence of at least one of the four harmful practices associated with caregiver management of diarrhea in children under the age of five, regardless of study design or representativeness of the sample population. Qualitative articles, or quantitative articles not meeting the quantitative inclusion criteria, were included if they presented substantive findings on beliefs, motivations, or context related to at least one of the four practices in caregiver management of childhood diarrhea. Publications were excluded if they exclusively reported data collected prior to 1990, exclusively reported provider practices, reported findings post-intervention only, or did not specifically focus on treatment of children under 5 years of age. Due to the variety of study designs included in the review, study quality was not formally assessed, because multiple quality assessment frameworks would have been required.
Data extraction was completed by the first author (EC). For all studies, information on the study design, study population, and sample size was extracted. For studies reporting prevalence of practices, data were extracted on the definition of the practice measure, the reported prevalence of the practice, and variation in the practice by other factors (reported as stratified prevalence or odds ratio). For non-prevalence studies, data were extracted related to beliefs, motivations, or context directly related to one or more of the harmful practices and then classified by common themes.
We summarize the results for each of the four harmful practices in the results section of the manuscript. For each practice, we: (1) describe how the practice was defined and measured in these studies; (2) summarize reported findings on prevalence, including variations by characteristics of the diarrhea episode, child, caregiver, and household; and (3) report on beliefs, motivations, and contextual factors investigated and relevant results.

Results
The initial search yielded 2,266 articles in Pubmed, 2,512 articles in Embase, 1,512 articles in Ovid Global Health, and 1,890 articles in the WHO Global Health Library. After removing duplicates, 4,270 unique articles remained. Title and abstract review and full article review were conducted by the first author (EC). After reviewing titles and abstracts, 294 articles were identified for full article review. Based on a review of the full article, 157 articles did not meet the inclusion criteria and a full text copy of 23 manuscripts could not be located. In total, 114 publications met the inclusion criteria and were included in the review (Fig. 1). Of the 79 studies reporting the prevalence of at least one harmful practice, 54 studies utilized a population-based cross-sectional sample (3 nationally representative), 12 studies used a non-cross-sectional design but included a representative population sample, and 13 studies employed a nonrepresentative sample. Of the 35 studies reporting on beliefs, motivations, or context for harmful practices, 9 studies used exclusively qualitative methods, 8 studies used mixed-methods, and 18 studies used exclusively quantitative methods (12 with a representative sample, 6 with a non-representative sample). Although there have been summaries of relevant Demographic and Health Survey (DHS) and Multiple Indicator Cluster Survey Fig. 1 Flow of studies considered in the systematic review (MICS) findings [5,6], we were unable to identify any country-specific secondary analyses on this topic.

Study characteristics
The publication dates of the 114 studies included in the review were relatively evenly distributed over the period from 1990 to 2013, with publications clustering slightly in the early 1990s and late 2000s/early 2010s. The majority of studies were conducted in South Asia and sub-Saharan Africa (Fig. 2). The number of publications reporting on the prevalence of each of the four practices varied, with the highest proportion reporting on inappropriate medication use (70 %), followed in order of frequency by food restriction (56 %), curtailment of fluids other than breast milk (53 %), and breastfeeding restriction (37 %).
Respondents in the majority of prevalence studies were caregivers of children under 5 years of age, although some studies interviewed mothers exclusively. The age of children referenced for the practice also varied, with the majority of studies referencing children under 5 years of age. The definition of the diarrhea reference episode also varied, ranging from diarrhea in the past 24 h to the most recent diarrhea event, although the most common reference period was the previous two weeks.

Fluid curtailment
The measurement of fluid intake, and prevalence estimates, varied widely across studies (Table 1, Column 4). Many studies differed in their definition or failed to specify if fluid restriction included or excluded breastfeeding or assessed amount of fluid offered versus consumed. The reported practice of curtailing fluids during a recent episode of diarrhea ranged from as low as 11 % of caregivers in Mirzapur, Bangladesh [7] to over 80 % of caregivers in Kenya's Nyanza province [8]. Where specified by the study authors, the practice of stopping all fluids was uncommon, generally reported in fewer than 10 % of episodes.
Multiple studies explored variations in fluid curtailment by characteristics of the diarrhea episode, child, caregiver, and household (Table 2). Fluid curtailment was associated with diarrhea severity and vomiting in two studies [9,10], whereas increase in fluid was associated with long illness duration and poor appetite [11]. Studies in Pakistan, Bangladesh, and Saudi Arabia found no clear association between fluid restriction and the age of the child [12][13][14]. However, a study in Mozambique reported that less fluid was given to infants relative to older children [15]. Younger mothers and mothers who did not work outside the home [12] and less educated mothers [16] were more likely to curtail fluids.

Breastfeeding reduction
Many studies reported the practice of breastfeeding reduction or cessation during diarrhea episodes (Table 1, Column 5). Most studies found that among mothers breastfeeding their child prior to the onset of diarrhea, fewer than 10 % of mothers stopped breastfeeding during the episode. The practice of breastfeeding cessation ranged from no mothers reporting breastfeeding cessation in a surveillance study in northeast Thailand to 62 % of mothers reporting stopping breast or milk feeding in a hospital-based study in Saudi Arabia [20,28]. The practice of breastfeeding cessation was higher in hospital samples compared to samples from the general population. Where breastfeeding reduction was reported, on average one quarter of mothers reported reducing breastfeeding, although there was significant variation in the practice.
Multiple studies assessed variance in breastfeeding restriction by factors including characteristics of the diarrhea episode, child, caregiver, and household ( Table 2). One study found younger and less educated mothers were more likely to reduce breastfeeding during episodes of diarrhea [12].
Mothers reported ceasing or reducing breastfeeding when their child had diarrhea for various reasons (Table 3). Mothers reported stopping or reducing breastfeeding because of beliefs that breastmilk was too fatty to be digested [20]. Others reported continued breastfeeding would not reduce the duration of diarrhea [20,29] or could cause or worsen the diarrhea [18,19,29]. Caregivers in two studies believed specific types of diarrhea must be treated with breastfeeding cessation [29,30]. In multiple cultures, "dirty" breast milk or secretion of ingested food through breast milk was thought to cause certain types of diarrhea. Mothers received treatment or a modified diet to improve the quality of their breast milk [31][32][33][34] or children were weaned [35]. Some caregivers stated they were following the advice of healthcare providers by restricting breastfeeding [20,36]. Older relatives were also important sources of information on feeding practices during diarrhea episodes [25,31]. In some studies, mothers continued feeding but diluted milk or formula [29], switched to powdered or goat's milk [37], or only gave water [38].

Food restriction
The measurement of food restriction, and prevalence estimates, varied widely across studies (Table 1, Column 6). Many studies differed in their definition or failed to specify if food restriction was measured only among those eating solid foods prior to illness, whether breastfeeding was included or excluded, and whether amount of food offered versus consumed was measured. Findings on restriction of specific foods have been included for context but not in prevalence estimates of overall food restriction ( Table 1). The practice of stopping all food ranged from as low as 3 % of mothers stating they stopped giving solid or semi-solid foods during the episode in Oyo State, Nigeria [26] to as high as 53 % of mothers reporting they stopped feeding in Kenya [39]. As expected, measures that included the reduction of feeding in addition to complete restriction of feeding showed higher rates of food restriction, mostly within the range of 30-60 % of episodes.
Multiple studies addressed the variance of food restriction by other factors, including characteristics of the diarrhea episode, child, caregiver, and household (Table 2). Food curtailment was associated with dehydration and more severe disease [40], seeking care outside of the home, and ORS use [41]. In one study, caregivers were more likely to withhold food if a child had fever or a low appetite [11]. Another study found children less than 2 years of age were more likely to receive continued feeding compared to older children [42]. Two studies found that less educated mothers were more likely to restrict foods [12,16].
Motivation for food restriction differed (Table 3). Some caregivers reported that a child's diet should be restricted because of beliefs that a child cannot eat or digest as much during a diarrhea episode [22,43] and feeding can exacerbate or prolong diarrhea episodes [19,22,29,[44][45][46]. Belief that only certain foods should be restricted because they can aggravate diarrhea was common across countries and included a range of foods such as meat, milk, sweet food, greasy food, high Table 2 Factors associated with harmful practice (Continued)

Child
Older child age [13,14] Caregiver Higher maternal education [64] [ 60] Household Greater household income [60,87] Live in urban area (vs rural) [93] [29,37,38,43,[47][48][49][50][51][52][53][54]. Alternatively, in two studies some caregivers reported that specific foods were customary and should be given during a diarrhea episode to strengthen the bowel or soothe the stomach [36,52]. Some caregivers reported that restriction of certain foods was based on long held folk tradition [29,47]. Others reported that diet alteration is based on the type or perceived cause of the diarrhea [18,29,55]. Elderly relatives, neighbors, and health care providers were reported to influence mothers' feeding practices in many contexts [22,23,25,27,29,36,53,56,57]. Some caregivers reported that a child's diet was not restricted during diarrhea because it was already limited [27,44,58]. One study reported mothers coaxed their child to eat more [36], but others reported some mothers of children with decreased appetite were unfamiliar with encouraging children to eat [22,44] or had little time to prepare additional food because they were caring for the child [22]. One study suggested caregivers felt continued feeding was less important if they had been given some treatment at a health facility [31].

Inappropriate medication use
Many studies reported the use of drugs to treat diarrhea in children under five (Table 1, Column 7). The most commonly reported measures were the use of an antibiotic or antimicrobial, followed by use of any medicine, and the use of an antidiarrheal or antimotility agent. While antibiotics are recommended for treatment of dysentery or cholera, most studies did not differentiate between simple and dysenteric diarrhea when reporting on antibiotic use. The Lives Saved Tool (LiST) attributes 7 % of diarrhea cases in children under 5 to dysentery [59], therefor it may be inferred that high antibiotic use rates are inclusive of inappropriate antibiotic use. A hospital-based study in Enugu, Nigeria highlights the difficultly of collecting information on the type of medicine used to treat diarrhea. The study reported that 70 % of mothers misclassified antibiotics and analgesics as antimotility agents when self-reporting drugs used in diarrhea treatment [60]. Multiple studies outside of this review have shown that the accuracy of drug recall varies by questionnaire design and method of assessment [61]. Reported use of antidiarrheal and antimotility agents was generally lower than reported use of antibiotics. Use of antibiotics at any point in an episode ranged from 10-77 %. Antidiarrheal use ranged from 3-45 % of diarrhea episodes, with the exception of very high reported use (74 %) in Egypt in 2002 [62]. Use of any drug for a diarrhea episode occurring in the previous 2 weeks ranged from 26-76 %. Studies that used a shorter reference period limited to the previous 24 h reported lower rates of drug use at around 20 %.
Multiple studies addressed variance in inappropriate medication use by factors including characteristics of the diarrhea episode, child, caregiver, and household (Table 2). A hospital-based study in Nigeria found children who had received an antibacterial or antidiarrheal at home presented to the hospital with more severe dehydration than those children who did not receive these drugs [60]. Antibiotic and/or antidiarrheal use were associated with seeking care outside of the home [11,41] and use of ORT [60,63]. Two studies in Enugu, Nigeria reported conflicting associations between maternal education and antibiotic use [60,64].
Caregivers reported using antibiotics and other drugs to treat diarrhea because they were accessible and believed to be efficacious (Table 3). Multiple studies reported caregiver beliefs that modern medicines are powerful [64][65][66][67], and more effective in treating diarrhea than ORS [65,68]. Multiple studies reported drugs were widely available and affordable in the public and private sector, typically without prescription [35,38,40,44,49,52,64,69]. In many contexts, caregivers stocked drugs at home, purchasing them in advance or saving leftover medication from previous illnesses [33,37,38,52,70]. Caregivers perceived drugs to be cheaper and more accessible than ORS, particularly given the flexibility to purchase a few tablets for little money [64,65,71]. Use of antibiotics in the treatment of pediatric diarrhea has become routine for both health care providers and caregivers in some contexts [18,40,66]. Caregivers may have also influenced provider behavior as caregivers' preference for drug therapies creates pressure on providers to give medications in addition or instead of ORS [28,33,65,72]. Drugs were given in sub-clinical doses in multiple studies [67,69,73]. It was common in studies for children to receive multiple drugs for a single episode of diarrhea, often from the same source [67,[74][75][76][77]. A study in Brazil found drugs were used more commonly to treat episodes of longer duration [63], although initial treatment of diarrhea at home with drugs was common in a study in Mali [78]. Multiple studies suggested treatment with modern medicines may be related to the perceived cause or type of diarrhea [18,52,60,[79][80][81]. Treatment seeking was often related to inappropriate use of medicine for diarrhea management [33,57,62,82].

Discussion
This is the first review, to our knowledge, that addresses harmful practices related to fluids, feeding and medication use during episodes of childhood diarrhea. The findings indicate that there have been many studiesboth quantitative and qualitativethat have documented these harmful practices. However, reported prevalence varies greatly across study populations, and we were unable to identify clearly defined patterns across regions, countries, or time periods. A limited number of studies looked at the variation of these harmful practices across potential influencing factors, including characteristics of the diarrhea episode and child, caregiver, or household-level traits. Findings of association differed across studies.
The motivation for harmful practices during diarrhea treatment also appears to vary across populations, although studies consistently report general caregiver concern for their child's health and caregiver action to treat the illness to the best of their knowledge and abilities. Caregivers reported that their actions were based on the advice of health care providers, community members, or elderly relatives, as well as their own observations or understanding of the efficacy of certain treatments for diarrhea. Others reported following traditionally held beliefs on the causes and cures for specific diarrheal diseases.
Across studies, the measurement of harmful practices was inconsistent and not guided by a conceptual or theoretical framework. Most studies were focused on general practices in diarrhea treatment, and harmful practices were rarely a primary outcome of interest. This has limited the availability and quality of data on the topic. Variations in study design, sample populations, diarrhea episode reference periods, and measurement definitions make drawing comparisons and conclusions across studies challenging. This is further compounded by inconsistent quality in data collection and reporting. Most studies relied on sub-national population samples and many were limited to small sample sizes. The variation in treatment practices by perceived type of diarrhea highlights the importance of using local terminology in order to capture all episodes of diarrhea as perceived by the community [83]. Although the majority of studies included in this review used a recall period of diarrhea in the past two weeks, there was some variation ranging from the past 24 h to past six months or the "most recent" episode of diarrhea. Fischer-Walker and her colleagues highlight the importance of using a shorter recall period for capturing episodes of diarrhea of varying severity [83].
Although this systematic review highlighted limitations of existing research, the available evidence suggests that harmful practices in diarrhea treatment are common in certain populations. A multicountry analysis using MICS data from 28 countries between 2005-2007 reported the majority of mothers did not maintain their child's nutritional intake during illness [5]. Analysis of DHS data from 14 countries between 1986-2003 suggests a decreasing trend in continued feeding in a majority of countries [6]. These practices can reduce correct management of diarrheal disease in children and result in treatment failure and sustained nutritional deficits. The lack of consistency in sampling, measurement, and reporting identified in this literature review highlights the need to document harmful practices using standard methods of measurement and reporting. Going forward, studies in this area would benefit from the development and use of a broader conceptual framework to ensure that the research is theory-driven and regularly synthesized. Multi-country analyses using MICS and DHS data have been conducted in the past, but they have tended to focus on positive treatment practices rather than harmful practices [5,6]. Assessing harmful practices with nationally representative data and standardized measurements, through the analysis of the most recently available DHS and MICS data, can contribute to the discussion on improved care of diarrheal disease in children under five.
The strengths of this literature review include applying a systematic process for searching and summarizing the literature, and accessing articles during a time frame in which global efforts focused on improving coverage. This review was limited by the inclusion of only peerreviewed literature and the exclusion of non-English language publications. Additionally, the quality of individual articles was not assessed, allowing for the potential inclusion of studies with misrepresentative findings.

Conclusions
Harmful practices in the management of childhood diarrhea are prevalent to varying degrees across cultures and include fluid and breastfeeding curtailment, food restriction, and inappropriate medication use. Inappropriate management of diarrhea episodes can result in higher risk of mortality through increased levels of dehydration or lasting health consequences as a result of nutritional restrictions or prolonged diarrhea illness. These practices must therefore be addressed as a matter of urgency in maternal, newborn and child health programs. These programs need to target not only the behaviors of child caregivers, but the broader social network, because our findings show that these practices are often informed by traditional beliefs, popular knowledge, and the instruction of authority figures, including elderly community members and health workers. Broader health systems interventions are also needed to address the alarming findings of high rates of inappropriate use of medications during diarrhea episodes. In addition, the global health community must do a better job or measuring the prevalence of these practices in standard ways, to produce evidence that can be used as the basis for action.

Additional file
Additional file 1: PubMed Search Terms. (PDF 68 kb)