Improving nutrition information systems: lessons from Kenya

Location: Kenya

What we know:  Strong nutrition information systems (NIS) are critical for timely, effective nutrition emergency response.

What this article adds:  Key lessons post-2008 emergency in Kenya prompted a concerted effort to ensure quality nutrition information is available to guide programme response. Government-led technical forums were established, including the National Nutrition Information Technical Working Group (NITWG). Challenges identified included non-standard indicator use, duplicative and inconsistent data collection mechanisms, and inconsistent results from infant and young child feeding (IYCF) data. Recommendations from a detailed evaluation by the US Centers for Disease Control and Prevention (CDC) were adopted and implemented. These included streamlining indicators and processes in district NIS, aligning anthropometric data with global guidance, capacity development of the drought early warning surveillance system, standardising indicators and survey schedules, and improving data quality of large-scale surveys. Three data clinics were held to critically review and triangulate data, standardise tools and methods and provide stakeholder guidance. A user-friendly, accessible, web-based central data database was developed. NIS developments improved timely response in 2016 to drought-related emergency. Technical support and capacity accessed and coordinated through the NITWG was instrumental to a successful Kenya Demographic Health Survey (DHS) in 2014. Key ingredients of success were strong government leadership and stakeholder commitment, a strong coordination mechanism, inter-sector linkages, mapping of partner capacity, integration within existing systems, and regular capacity-strengthening. A committed team and structures to follow through recommendations and actions is key.

• Indicator definitions and collection methods used were not standardised despite global guidance, especially for non-anthropometric indicators such as infant and young child feeding (IYCF), food security, micronutrient deficiencies and water and sanitation. This severely limited trend/meta-analysis across regions.
• Some monthly health facility indicators reported in the District Health Information Systems (DHIS), such as on breastfeeding and stunting, were difficult to interpret. The stunting indicator from health facilities significantly underestimated the burden of stunting, mainly due to low reporting rates, limited access to equipment, poor quality of height measurements due to limited staff, and age bias (younger children visit facilities more frequently). The indicator on exclusive breastfeeding (EBF) overestimated EBF rates due to self-reporting bias; data were only collected from mothers who visited the health facility and were therefore not representative of the population and there was likelihood of the same mother responding multiple times, depending on how often she visited the clinic. When these indicators were compared to representative, population-based surveys, the results from health facilities were always significantly different, making it difficult to inform programmes on progress.
• IYCF indicators integrated into SMART surveys had wide confidence intervals, erratic results and were difficult to compare over time.

To address these concerns, in 2013 the NITWG called for an evaluation of the Kenya nutrition surveillance system by the US Centres for Disease Control and Prevention (CDC) with support from UNICEF. The team from CDC Atlanta² evaluated NIS and made several recommendations to improve the system. One of the key recommendations was to support government-led information systems that already routinely collect nutrition data (the DHIS, the National Drought Management Authority (NDMA) early warning system and the national surveys conducted by the KNBS) because they have the widest coverage, enjoy government leadership and ownership (therefore constituting a more sustainable approach) and are more likely to be supported by multiple partners. Additional recommendations focused on data quality improvement initiatives across all nutrition information sources and on strengthening inter-sector linkages. The NITWG adopted the evaluation report and prioritised planning and implementation of actions related to improving data quality and reporting across all nutrition information sources. This article shares the experiences of the NITWG in delivering on this remit.

Review of the Nutrition Information System

The regular, twice-monthly NITWG meetings provided inadequate time to address the evaluation recommendations, so a ‘data clinic’ workshop was held 3-5 September 2013 to review with key partners the indicators, methodologies and timing/frequency of surveys and assessments. This was led by the MoH with technical and financial support from UNICEF and Action Against Hunger (NITWG chair); participants included the World Food Programme (WFP), Food and Agriculture Organization (FAO), NGO partners, NDMA and food security technical experts. To date there have been three data clinics each tackling key issues of indicator definition, quality, reporting and methodologies (see later), and reviewing the progress and challenges observed since the previous clinic. The following summarises the evaluation findings and recommendations and subsequent actions taken.  

District Health Information Systems

The CDC evaluation reviewed 11 key nutrition indicators routinely collected through the DHIS system: underweight, stunting, vitamin A supplementation, iron folate supplementation, treatment of severe acute malnutrition (SAM), treatment of moderate acute malnutrition (MAM), deworming, growth monitoring, early breastfeeding initiation, exclusive breastfeeding and micronutrient powder supplementation. Key challenges highlighted were as follows:

The same indicator was collected in more than one form (e.g. MoH 711 for Child Health and MoH 713 Nutrition Monthly Monitoring) and used different age categorisations, classifications and reporting rates.

Health facility stunting data were significantly different from survey data. An estimated 10% of the facilities³ had information on stunting at the time of the review; low reporting/poor data quality was mainly linked to limited access to height boards, lack of time and staff for measurement, and skewed age distribution of children presenting at the health centre (majority under one year of age).

Key recommendations and actions taken to address these issues were:

• Standardise reporting where there was more than one form collecting the same information. It was agreed that all nutrition indicators will be reported in one form. To help ensure health facilities had adequate supplies of the required DHIS tools, printing and distribution were supported by UNICEF and WFP.
• Two indicators – stunting and exclusive breastfeeding rate – cannot be accurately reported using the DHIS; these indicators should be collected in population-based surveys only and the NITWG should work closely with the KNBS to ensure accurate collection (including training and developing systems for quality checks on indicators). The NITWG has since offered committed technical support and guidance to ensure quality of nutrition indicator measurements in large-scale surveys conducted by the KNBS, such as the 2014 Demographic Health Survey (DHS) and the Kenya Integrated Household Budget Survey (KIHBS).
• The NITWG should work closely with counties to support the capacity-building of staff to improve the analysis and quality of data and provide guidance to staff on simple data quality checks, discussing findings in the in-charge facility meetings.  Through the support of UNICEF, a nutrition information officer was made available to visit the county health facilities and provide on-the-job training, guidance and support on nutrition indicators in the Health Management Information System (HMIS) system. 

National Drought Management Authority early warning surveillance system   

The NDMA collects monthly information from sentinel sites in the arid and semi-arid areas (ASAL) of the country. Indicators include food security indicators such as rainfall, vegetation coverage index (VCI), market prices (food and livestock), livestock body condition, water availability and cost, milk production and availability, mid-upper arm circumference (MUAC) in children aged 12-59 months, and health and morbidity data. The evaluation found that MUAC data were reported as the prevalence of children with MUAC < 135 mm, which was referred to as children at risk of acute malnutrition, and not further disaggregated in the categories (1) <135 mm to ≥125mm (at risk), (2) <125 to > 115 mm (MAM) and (3) <115mm (severe acute malnutrition (SAM)) as internationally accepted. It was agreed that applying the standard thresholds for acute malnutrition and improving data quality would help realise the potential of the NDMA early warning system.

To address this, the NITWG linked with the NDMA to review the current data and discuss avenues of improving data quality through joint data quality assurance visits and agreed on capacity-building plans for NDMA information officers. In 2013, UNICEF supported a two-day training for country nutrition officers on nutrition information analysis and reporting.  Focus areas were proper measurement technique for MUAC; nutrition indicator analysis, including DHIS information review; and the use of emergency nutrition assessment software (ENA) for SMART to check the quality of MUAC data sets.

In late 2014 and 2015, with the support of the UNICEF regional office, the NDMA information officers from all ASAL areas were trained on SMART surveys. This enabled them to participate and take ownership of nutritional surveys in their counties and be better placed to report on county nutrition surveillance. The Kenya Food Security Steering Group (KFSSG) Data and Information Subcommittee members and the chairperson of the KFSSG were also trained on SMART and nutrition surveillance methods to better interpret nutrition information. This training was crucial to help the KFSSG key members understand the rigour and processes involved in nutrition surveys and how this information can be used for better programming.

On review, the NDMA early warning system, in consultation with the NIWG, agreed to report the number of acutely malnourished children using the global standards of <125mm and 115mm. For purposes of trends and continuity, children less than <135 mm continued to be reported.

Standardising indicators and timing of nutrition surveys

Surveys that collect nutrition information in Kenya included SMART surveys at a county level in the most affected locations, DHS and special nutrition studies and research. To track progress and allow development of a central database of raw data and reports, standardised core nutrition indicators and variables were agreed; during data clinics, minimum indicators and required data were determined for each sector and for maternal and infant and young child nutrition (MIYCN) assessments. Guidance on key technical issues was provided: IYCF surveys should no longer be combined with SMART surveys (since the SMART sample size for children under two years of age is too small for IYCF indicator data collection and IYCF behavioural changes are likely to happen over a longer period, requiring less frequent surveying than is typical for SMART surveys). A coordinated and standardised survey schedule was also agreed. Surveys should be conducted at the start of the hunger season and aligned to food security assessments to allow for complementary analyses. A detailed survey plan, based on consultations with the KFSSG and DISK members, was developed for the drought-prone areas to ensure timely nutrition information collection to feed into overall food security and nutrition seasonal assessments.

These developments enabled the NITWG to more easily conduct Integrated Phase Classification (IPC) analysis for food security and nutrition in most ASAL areas (less information was available for semi-arid parts of the country where surveys were less frequent). For example, the existence of a central database ensured that all data to inform the analysis were available and clearly organised, allowing all partners to review information. This helped the NITWG to monitor data quality and availability across the ASAL areas and work closely with NDMA to improve data quality. In addition, seasonal assessment findings and IPC analysis were released at the Kenya Food Security Meeting (KFSM), a high-level food security and nutrition forum, ensuring that nutrition information was shared with key stakeholders in the country for immediate action. IPC analysis products (IPC maps showing the most affected areas (see Figure 1 for examples), the caseloads of children affected with malnutrition and key recommendations) were shared in a one-page infographic. This quickly highlighted the nutrition situation, the number affected and the immediate assistance required. Released alongside the food security analysis, this report forms the basis for the sector emergency response plan that is produced both at national and county level, which is used to plan for interventions and mobilise required resources. An example of how timely, integrated assessment has informed emergency response is shared in Box 1.