Nutrition Training Improves Health Workers’ Nutrition Knowledge and Competence to Manage Child Undernutrition: A Systematic Review

Background: Medical and nursing education lack adequate practical nutrition training to fit the clinical reality that health workers face in their practices. Such a deficit creates health workers with poor nutrition knowledge and child undernutrition management practices. In-service nutrition training can help to fill this gap. However, no systematic review has examined its collective effectiveness. We thus conducted this study to examine the effectiveness of in-service nutrition training on health workers’ nutrition knowledge, counseling skills, and child undernutrition management practices. Methods: We conducted a literature search on nutrition interventions from PubMed/MEDLINE, CINAHL, EMBASE, ISI Web of Knowledge, and World Health Organization regional databases. The outcome variables were nutrition knowledge, nutrition-counseling skills, and undernutrition management practices of health workers. Due to heterogeneity, we conducted only descriptive analyses. Results: Out of 3910 retrieved articles, 25 were selected as eligible for the final analysis. A total of 18 studies evaluated health workers’ nutrition knowledge and showed improvement after training. A total of 12 studies with nutrition counseling as the outcome variable also showed improvement among the trained health workers. Sixteen studies evaluated health workers’ child undernutrition management practices. In all such studies, child undernutrition management practices and competence of health workers improved after the nutrition training intervention. Conclusion: In-service nutrition training improves quality of health workers by rendering them more knowledge and competence to manage nutrition-related conditions, especially child undernutrition. In-service nutrition training interventions can help to fill the gap created by the lack of adequate nutrition training in the existing medical and nursing education system. In this way, steps can be taken toward improving the overall nutritional status of the child population.


BACKGROUND
Child undernutrition can be reduced if health workers with adequate nutrition knowledge provide correct, adequate, and frequent nutrition advise to caregivers (1,2). Across the globe, the quality of health workers' nutrition knowledge -and, by extension, their counseling skills -has been a concern (3-7). Historically, medical training has lacked adequate and updated nutrition training that is in keeping with the situation and needs on the ground (8)(9)(10). As a result, health workers produced from teaching institutions have lacked adequate nutrition knowledge (3,6). Such health workers may also lack the competence and skills to provide basic nutrition advice to their clients (2,11). This incompetence, in turn, may be a factor deterring health workers from providing nutrition advice and management to their clients (12).
In practical terms, the process by which the knowledge acquired through nutrition training is translated into management practices may not be linear. However, the outcome of nutrition www.frontiersin.org training can be explained using the conceptual framework of general behavioral theories. Based on the Health Belief Model (27), for example, knowledge or education on a perceived threat or disease is likely to influence behavior change. Untrained health workers may feel incompetent to provide counseling to their clients even when they know or perceive the threat caused by a nutrition problem. When such health workers are trained to recognize threats posed by nutrition-related problems, they are likely to provide appropriate care. In an ideal situation, knowledge can impact practical skills; it can thus change health workers' behaviors.
Changes in health workers' nutrition-counseling behavior can be sustained even after training ends. This may be due to the rewards they may receive in the positive results of improving child feeding practices and nutritional status. According to Bandura's social learning theory, a change can be influenced by being rewarded or punished as a result of one's actions (27). Health workers' counseling behavior is more likely to last or recur if the nutrition knowledge received is used to counsel or manage a child with undernutrition so as to yield obvious improvements in the child's nutritional status.
Evidence is available for the impact of in-service nutrition training on health workers' nutrition knowledge, nutritioncounseling skills, and management of child undernutrition. However, no systematic review has summarized such evidence toward effecting policy change. A few review articles have demonstrated the importance of nutrition training for health workers (7). Other reviews have shown the effect of health workers' counseling of caregivers on feeding practices including dietary diversity, feeding frequency, and energy intake (28), complementary feeding, and children's nutritional status (1,29,30). However, we could not find any systematic review on the impact of nutrition training for health workers on their own knowledge, attitudes, and child undernutrition management practices. Therefore the objective of this systematic review was to evaluate the impact of nutrition training interventions among health workers on their nutrition knowledge, nutrition counseling, and child undernutrition management practices.

METHODS
The population, intervention, comparator, outcome (PICO) question to be addressed in this study was framed as follows: what is the effectiveness of nutrition training of health workers on their nutrition knowledge, nutrition counseling, and undernutrition management practices among children at risk of or suffering from undernutrition as compared to those who did not receive such training? For this review, we included studies with nutrition training interventions.
We defined nutrition training for health workers as any form of in-service training given to health workers and designed for continuing professional development (CPD), continuous medical education (CME), research purposes, or as part of a health project or program. We used a World Health Organization (WHO) and International Labor Organization (ILO) definitions of health worker to select health cadres as the population of interest (31,32). Such health cadres included doctors, nurses, midwives, mid-level providers, dieticians, nutritionists, and pharmacists.
Three outcome variables were assessed in this review: health worker's nutrition knowledge, nutrition counseling and/or general counseling skills, and management skills for child undernutrition. Health worker's nutrition knowledge was measured using a standard scale or pre-made set of questions to test knowledge specific to the training given. We regarded nutrition counseling as any specific advice given by a health worker to caregivers on nutrition, feeding characteristics, dietary composition, or food intake. Such advice might have followed evaluation of the patient's nutritional status, feeding behavior, dietary composition, or training on poor nutrition conditions. Depending on the availability of the data, we included assessment of quality and frequency of counseling to assess both skills and counseling acts of trained health workers.
We defined management practices for undernutrition as activities health workers perform toward management of poor nutritional status. Depending on data availability, this might include assessment of undernutrition using anthropometric scores, assessment of micronutrient deficiency, treatment of associated conditions, treatment of undernutrition by prescription of supplements, or monitoring by growth charts, among other methods.
The protocol for this systematic review was registered at PROSPERO http://www.crd.york.ac.uk/PROSPERO on February 6, 2013. The registration number for this review is CDR42013003800. It is available at http://www.crd.york.ac.uk/ PROSPERO/display_record.asp?ID=CRD42013003800.

INCLUSION CRITERIA
Based on the nature of the intervention and the outcome of interest, we included studies with nutrition training interventions. As for the study design, we included studies conducted as RCTs, cluster RCTs, quasi-experimental studies, and pre-post-intervention longitudinal studies with or without comparison groups.

EXCLUSION CRITERIA
We excluded any study where in the structure and/or quality of the training intervention provided to health workers was unclear from the description provided. We also excluded studies involving cadres of health workers outside the scope of the WHO that included community health volunteers, medical and nursing students, medical interns, and peers trained to provide specified health services. Non-interventional studies were also excluded from this review. This included cross-sectional studies, case reports, and non-interventional qualitative studies.

DATA SOURCES FOR EXISTING REVIEWS
We first searched for similar reviews or registered review protocols to avoid duplication and redundancy according to the standard review guidelines of the Center for Review and Dissemination (CRD) and Cochrane (33,34). To this end, we searched such protocols listed in the Cochrane library and Cochrane Database of Systematic Reviews (CDSR). We used a similar approach to search other important databases for systematic reviews, including the Database of Abstracts of Reviews of Effects (DARE), the Educational Resources and Information Center (ERIC), the Campbell Library of Systematic Reviews, and the National Institute for Health and Clinical Excellence (NICE). Two independent researchers conducted the search for existing protocols and similar review articles.

EVIDENCE SEARCH STRATEGY
Two independent researchers conducted a literature search based on the published review protocol. The search was conducted in five medical databases: PubMed/Medline, CINAHL, EMBASE, ISI Web of Knowledge, and WHO regional databases. The search was limited to a 15-year-publication period (1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012), to ensure that we obtain enough evidence. Also, most of standard nutrition training interventions became more common after the publication of Integrated Management of Childhood Illnesses (IMCI) by WHO in 1997. For the PubMed/Medline database, we followed the search strategy outlined in the CDR register. Similar key words were then used to conduct searches in other selected databases. In order to ensure that we captured most of the relevant articles, we also conducted a hand search using references from key identified articles and archives of a journal with similar specialty -the Journal of Human Resources for Health. Figure 1 shows the results of searching and data management procedures according to the PRISMA check list ( Table 1) (35).
A total of 3910 articles were retrieved from various sources. Of these, 646 studies were from PubMed/Medline. We retrieved a total of 3255 from other databases: 341 from CINAHL, 1543 from EMBASE, 1249 from ISI Web of Knowledge, and 122 from WHO Regional databases. A total of nine studies were additionally obtained based on hand searches and review of specific journal databases.
On an initial screening of articles, we excluded a total of 3865 abstracts and articles due to duplication between PubMed/Medline and other databases and those, which were irrelevant to the research question. Of the 45 studies remaining, we further excluded 20 studies following a full-text assessment of eligibility. The reasons for exclusion were as follows: different outcome variable to that of interest (n = 10), unclear or different intervention (n = 5), unclear or different population (n = 3), and different study design (n = 2).

ANALYSIS STRATEGY
Also shown in Figure 1 is the distribution of the 25 studies included for final analysis according to outcome variables. A single study could incorporate more than one outcome variable of interest. A total of 18 studies had health workers' nutrition knowledge as an
The duration of the intervention, population characteristics, and measurement of outcome variables differed across each of the included studies. Study design also differed from one study to another though they all aimed generally at a nutrition training intervention for health workers ( Table 4). To avoid the obvious risk of heterogeneity, we did not conduct a meta-analysis. Instead, we sought to write a narrative summary to describe our results and stratified such description according to the outcome of interest.

RISK OF BIAS
We assessed risk of bias (RoB) for each cluster RCT and nonrandomized studies (Tables 2 and 3). We used the Cochrane RoB tool (34), to examine five types of RoB for randomized trials. These included selection, performance, attrition, detection, and reporting biases. Of the six cluster RCTs, only one study (13), had a high risk of selection and performance bias.
To assess quality and RoB for non-randomized studies, we used the Downs and Black scoring system (50). This tool has 27 checklists and is used to examine RoB for observation studies. We used the tool to examine reporting bias, external validity, internal validity-bias, internal validity-(confounding) selection bias, and power of the study. Generally, the included non-randomized studies had good external validity and internal validity. However, 10 out of 18 of non-randomized studies included had low or unknown power to detect a clinical important effect of the intervention. This is due to the nature of the intervention and methods used to recruit their participants. Most of nutrition training used all available health workers, selected non-randomly, and without prior or unspecified sample size calculations.

GENERAL DESCRIPTION OF THE INCLUDED STUDIES
Out of 3910 retrieved studies, a total of 25 studies were included in this systematic review. These studies were conducted in countries from Africa, North and South America, Europe, Asia, and Australia ( Table 4). The selected studies were also conducted in varied socio-economic levels, from least to most developed countries. Among the included studies, six were cluster RCTs, one was a controlled non-randomized trial, 17 were pre-post-intervention evaluations, and one was a quasi-experimental design study. The included health worker cadres ranged from specialists in pediatric care to general practitioners (GPs), nurses, midwives and obstetricians, nutritionists, dieticians, and mid-level providers.

EFFECTIVENESS OF THE INTERVENTION TO IMPROVE HEALTH WORKERS' NUTRITION KNOWLEDGE (A NARRATIVE SUMMARY)
Based on local adaptation of IMCI feeding guidelines, the key feeding recommendations identified were as follows: increase breast and     high post-training nutrition knowledge compared to doctors in the control group. In the UK-based RCT among GPs, nutrition training did not significantly change health workers' nutrition knowledge (14). However, GPs in the intervention group were 30% more likely to believe that their nutrition knowledge was up-to-date compared to their counterparts in the control group (P = 0.001).
In an Italian controlled non-randomized trial with a comparison group (16), the mean nutrition knowledge score of nurses, midwives, and doctors increased from 0.41 to 0.72 in Group 1 and from 0.53 to 0.75 in Group 2 after the nutrition training. Compared to the delayed intervention control group, health workers in the intervention group registered a significant change in knowledge and skills (P < 0.01) in a quasi-experimental study conducted in the US.
Fourteen studies were conducted using a pre-post-intervention design in Australia (37), Denmark (17,18), Gambia (39), UK (41), US (20,21,49), Sweden (42), Ireland (44), Ethiopia (46), Nigeria (47), and Zambia (24). In all of these studies, health workers' nutrition knowledge increased after nutrition training. Table 6 shows the result of the 12 reviewed studies with nutrition counseling as an outcome variable following nutrition training of health workers. Of these studies, six were cluster RCTs and three were conducted in Brazil among doctors and pediatricians (13,15,19). Across all three studies, a significantly higher proportion of doctors in the intervention group had better post-training counseling skills and performed more nutrition counseling compared to the respective control group. Physicians undergoing the training intervention also showed higher mean communication skill scores compared to untrained physicians (P < 0.01) (19). In a Pakistani cluster RCT among lady health visitors (LHVs) working at a health facility as mid-level providers, 82% of participants in the intervention group registered improved post-training communication skills compared to 51% in the control group (P = 0.015). In this study, a higher proportion of trained LHVs reported increased counseling skills compared to their counterparts. GPs with nutrition training in the UK study (14), meanwhile, were 30% more likely to provide dietary advice that was completely appropriate (P = 0.01). In Peru (23), twice as many mothers in the intervention group received post-partum nutrition advice compared to their control group counterparts following the nutrition training intervention for their health workers (P = 0.02).

EFFECTIVENESS OF INTERVENTION TO IMPROVE NUTRITION-COUNSELING SKILLS OF HEALTH WORKERS (A NARRATIVE SUMMARY)
A total of six studies using pre-post-intervention evaluation of nutrition training of health workers reported nutrition-counseling skills as an outcome variable. These studies were conducted in the US (20,21,48), Ireland (44, 45), and Ethiopia (46). In all six studies, nutrition and general counseling skills of health workers improved after nutrition training. Table 7 summarizes the results of the 16 reviewed studies reporting management of undernutrition and management practices as an outcome variables following the nutrition training intervention. Within these studies, two of the three cluster RCTs were conducted in Brazil among medical doctors, pediatricians, and www.frontiersin.org         nutritionists (13,15). Doctors in the intervention group were more likely to report improved post-intervention practices in managing child undernutrition compared to their counterparts. In the Pakistani study, trained LHVs were more likely to plot children's weights, discuss appropriate foods with caregivers, and check mothers' understanding of imparted nutrition knowledge compared to their counterparts in the control group. In an Italian controlled non-randomized trial (16), all hospitals improved their compliance with WHO's "Ten Steps to Successful Breastfeeding" after undergoing the WHO baby-friendly hospital and counseling course. Nutrition and health professionals in the quasi-experimental design study in the US exhibited better nutrition management skills after the 6-week online nutrition training (49). Compared to health workers in the delayed intervention control group, those in the intervention group registered significant positive changes on knowledge and skills scores (P < 0.01).

EFFECTIVENESS OF INTERVENTION TO IMPROVE HEALTH WORKERS' MANAGEMENT PRACTICES FOR CHILD UNDERNUTRITION (A NARRATIVE SUMMARY)
Eleven pre-post-intervention studies on nutrition training of health workers were conducted in Australia (37), South Africa (38), Gambia (39), UK (40,41), Sweden (42), Denmark (43), Ireland (44, 45), Ethiopia (46), US, and Zambia (24). In all these studies, management practices and competence of health workers improved after the intervention compared to pre-nutrition training intervention levels.

DISCUSSION
This is the first systematic review to examine the effectiveness of in-service nutrition training to improve health workers' nutrition knowledge, nutrition counseling, and undernutrition management practices. In this review, we reviewed a total of 25 studies reporting on nutrition training interventions. Across all three of our outcome variables, significant post-intervention improvements were reported. First, in-service nutrition training improved health workers' nutrition knowledge. Second, the counseling skills and competence of health workers were also improved after in-service nutrition training. Third, the training intervention improved child undernutrition management practices of participating health workers. www.frontiersin.org Nutrition-counseling training using IMCI's "counsel the mother" module for five and a half days. It     A total of 18 studies, including five with a cluster RCT design, showed significant post-nutrition training improvements in health workers' nutrition knowledge. These studies were conducted in areas of varying social and economic levels and geographic characteristics. Health workers might also have been exposed to nutrition education during their college training (8). However, previous studies have indicated that such training is inadequate or not in keeping with the clinical reality encountered in practice (9,10). Lack of such knowledge might also cause them to refrain from providing nutrition counseling and care to their clients (8). Sometimes, due to lack of adequate nutrition knowledge, doctors feel that it is the duty of nurses or other cadres below them to provide nutrition counseling and care. To improve knowledge of such health workers, it is important to expose them to inservice nutrition training tailored to their environment, context, and health cadre (9). This will help to boost their competence and confidence in management of nutrition-related conditions including undernutrition.
In this systematic review, a total of 12 studies, including six cluster RCTs, showed a significant improvement in counseling skills among health workers with in-service nutrition training. Nutrition training would thus seem to be effective in improving health workers' nutrition knowledge. Nutritionally informed health workers may be more confident to address nutrition-related conditions in their patients (51). Such health workers may be better equipped to provide appropriate advice and counseling to their clients. The prevailing attitude toward nutrition counseling among medical doctors and pediatricians that such functions are not within their job description might also change after nutrition training. In this way, such trained health workers would be more likely to provide nutrition counseling (52). In line with the health belief model, nutrition knowledge provided to the health workers Frontiers in Public Health | Public Health Education and Promotion     through nutrition training is more likely to influence counseling behavior (27). Further, according to Bandura's social learning theory, such behavior or attitude change is mediated through cognitive processes and thus is learned through imitating and observing the actions of others (27). Accordingly, the reward that health workers can gain from their nutrition-counseling actions, such as better nutritional status or feeding practices in those they treat, may reinforce their counseling actions, thus making it a permanent habit. In this way, the quality of health workers with regard to nutrition counseling might be expected to improve. Our review showed that health workers' undernutrition management practices improved when they received in-service nutrition training. A total of 16 intervention studies, including three cluster RCTs, showed a significant improvement in management practices for child undernutrition after nutrition training. Barriers to effective management of child undernutrition include lack of nutrition knowledge and counseling skills among health providers (12). Such barriers can be ameliorated when health workers receive appropriate and tailored in-service nutrition training suited to their context and cadre.
Findings from this review should be carefully considered in the context of two primary limitations. First, the results are not based on meta-analysis to calculate the overall effect size of the intervention for each outcome variables. This was due to variations in the study designs and measurements used for outcome variables, and to differences in the competence, experience, and cadres of participating health workers. Such variations could have resulted in high heterogeneity. Hence, instead of meta-analyses, we explained each study separately in a narrative summary stratified by outcome variable. Although we did not pool our results, individual studies showed a significant effect of nutrition interventions on outcome variables.
Second, the included studies differed in the intervention's length and content. This might have caused differences in the measured outcome variables. However, most of the studies used standard nutrition training frameworks for health workers including the IMCI nutrition-counseling training module, breast-feeding www.frontiersin.org counseling training modules by both WHO and UNICEF, MUST training modules, and other comparable training based on formative research. Despite such differences, each study showed a significant improvement in one or more of the outcome variables.
Despite its limitation, findings thus presented may help decision makers to plan and conduct in-service nutrition training for health workers, an important building block toward building a strong foundation for any health system. This review is the continuation of series papers on the effectiveness of nutrition training of health workers. It is the first systematic review on the effectiveness of nutrition training for health workers on their nutrition knowledge, nutrition counseling, and undernutrition management skills. The other paper in the series found that, nutrition training of health workers improved feeding frequency, energy intake, and dietary diversity of children aged 6 months to 2 years (28).
In conclusion, in-service nutrition training of health workers improves their nutrition knowledge, nutrition and general counseling skills, and undernutrition management skills. Such nutrition training within the context of their practice is of paramount importance due to inadequate nutrition training in the health workers' mainstream medical and nursing education. In-service nutrition training can take different forms such as compulsory CME, nutrition seminars, workshops, or non-compulsory continuing professional education CPD. Whatever form it takes, nutrition training has the potential to improve the quality of health workers, making them more confident and competent in this key area and thus contributing to positive changes in population nutrition.

AUTHORS' CONTRIBUTIONS
Bruno F. Sunguya conceived the research questions, designed the study, participated in the literature review and analyses, and prepared the first draft. Krishna C. Poudel refined the research question and the first draft. Linda B. Mlunde contributed to the study design, participated in the literature review, and helped to prepare the first draft. David P. Urassa revised the protocol. Junko Yasuoka participated in the preparation of the first draft and revisions. Masamine Jimba reviewed the study protocol and manuscript, and approved the submission. All authors read and approved the final version of the manuscript for submission.